Predictive second cancer risk models based on peripheral neutron dosimetry of patients undergoing radiotherapy

FACULTAD DE MEDICINA DEPARTAMENTO DE FISIOLOGÍA MÉDICA Y BIOFÍSICA Grupo de Física Médica Predictive second cancer risk models based on peripheral neutron dosimetry of patients undergoing radiotherapy Leticia Irazola Rosales 16th September 2016 Francisco Sánchez Doblado. Catedrático del Departamento de Fisiología Médica y Biofísica, Facultad de Medicina de la Universidad de Sevilla y José Antonio Terrón Leon. Facultativo especialista de área del Hospital Universitario Virgen Macarena. Profesor Tutor Clínico de la Universidad de Sevilla. Certifican: Que Doña Leticia Irazola Rosales ha realizado bajo su dirección el trabajo de tesis “Predictive second cancer risk models based on peripheral neutron dosimetry of patients undergoing radiotherapy”, que presenta para optar al Grado de Doctor por la Universidad de Sevilla con Mención Internacional. Sevilla, a 10 de Junio de 2016. Fdo. Francisco Sánchez Doblado Fdo. José Antonio Terrón León A mis padres “Learn from yesterday, live for today, hope for tomorrow. The important thing is not to stop questioning” Albert Einstein Cover design Erika León Rosales “Je suis de ceux qui pensent que la science est d’une grande beauté. Un scientifique dans son laboratoire est non seulement un technicien : il est aussi un enfant placé devant des phénomènes naturels qui l’impressionnent comme des contes de fées. […] Telle est bien la beauté et la noblesse de la science : désir sans fin de repousser les frontières du savoir, de traquer les secrets de la matière et de la vie sans idée préconçue des conséquences éventuelles.” Marie Curie Agradecimientos El trabajo aquí presente no habría sido posible sin el esfuerzo del brillante grupo de científicos (y mejores personas si cabe) de las que me he rodeado durante esta etapa de mi vida. Por ello, mi agradecimiento a todos aquellos que se han cruzado en mi camino desde que comencé mi “aventura” en el mundo de la investigación, en lo cual tienen especial “culpa” mis compañeros del Centre Eugène Marquis, Rennes (gracias Jean Pierre y Caroline) y el Centro de Investigaciones BIomédicas de la Rioja (gracias Camilo). Me gustaría enfatizar mis agradecimientos a mis directores de tesis Paco y José Antonio, sin los cuales este sueño no hubiese sido posible. Paco gracias por tu confianza en mí desde el minuto cero, por tu apoyo incondicional y por contagiarme la pasión por este mundo. Me siento afortunada de haber tenido la oportunidad de aprender de la mano de alguien como tú. No solo me has formado en la investigación, sino que me has hecho crecer como persona desde tu cariño. José Antonio, gracias por todas esas horas compartidas, por todos los conocimientos inculcados y sobre todo, por despertar en mí el amor por el mundo clínico. Gracias por ser mi familia en Sevilla. No sería justa si no agradeciese a los principales colaboradores, sin los que este manuscrito nunca hubiese llegado a buen puerto…Beatriz, ha sido un honor trabajar a tu lado, gracias por tus brillantes ideas y maravillosas correcciones; Mónica gracias por aguantar nuestros “ataques” a tu persona y por las horas en el planificador; gracias Roberto por estar siempre dispuesto a ayudar e ilustrarme con tus inestimables aportaciones y Trini, gracias por tu incalculable disposición para echar siempre una mano. Mi eterno agradecimiento a la gente del Departamento de Fisiología Médica y Biofísica. A Loli por su constante preocupación por mi persona y a Fernando por ayudarme a sobrevivir a la burocracia. En especial, al grupo física médica: Elisa, Rita, José Antonio y Ana, gracias por todos los buenos momentos compartidos que me han hecho querer seguir adelante con esta “locura”. Gracias Antonio por tus siempre alentadores comentarios. Y por supuesto, a toda la gente del departamento por las horas compartidas en la “biblioteca”. No puedo ovidarme de todo el personal del hospital con los que he compartido los a veces tediosos, pero satisfactorios momentos de medidas, por su aguante y ayuda: Rafa, gracias por tu colaboración y disposición para la realización de los experimentos, sin los cuales este trabajo no existiría. Álvaro gracias por las críticas, los buenos consejos y mejores ratos. María y Héctor, mi eterna gratitud por todos los buenos momentos compartidos dentro y fuera. Y, por supuesto, gracias a mis queridos técnicos (especialmente aIsa M, Mari, María P, Pili y Juani) con las que tantas horas he compartido. Abstract ii important aspect that should be taken into account for second cancer risk estimations, being of special importance for children, whose life expectancy and radiosensibility are greater. For that, measurements with the new miniaturized active devices were performed for three different phantom sizes (child, teen and adult). Once these results were implemented, this new methodology was applied to 510 patients, starting the generation of an improved database that would allow a more patient specific analysis of second cancer risk, as a consequence of neutron contamination. Besides, the analytical peripheral photon model simultaneously developed by our group, has allowed the generation of a piece of software for the estimation of both peripheral doses. This has enabled a quick assessment of photon and neutron peripheral doses in clinical routine, from readily available parameters. Estimations of these doses were finally evaluated for some of the most common tumor locations, comparing conventional techniques and fractionations (3D-CRT, IMRT, VMAT) to newer ones (SBRT, FFF), regarding the three main linac manufacturers and energies (6, 10, and 15 MV). As a general pattern, hypofractionated modality, 10 MV photon energy and FFF irradiation mode have shown as the best alternatives in terms of peripheral dose reduction. Thus, a combined use of these options would imply a decrease of second cancer probability. Second cancer risk estimations could be easily performed from the here presented procedures by the direct use of the existing risk models, established by the international organisms (i.e. ICRP or BEIR). The universal methodology presented aims to provide an objective additional criterion (Second Cancer Probability, SCP), to be used in combination wih the previously existing radiobiological parameters as Tumor Control Prrobability (TCP) and Normal Tissue Complication Probability (NTCP), for the choice of the best radiotherapy strategy, thanks to its easy implementation in Treatment Planning Systems. List of publications as a consequence of the thesis iii List of publications as a consequence of the thesis Full papers as first author Irazola L, Lorenzoli M, Bedogni R, Pola A, Terrón JA, Sánchez-Nieto B, Expósito MR, Lagares JI, Sansaloni F and Sánchez-Doblado F. A new online detector for estimation of peripheral neutron equivalent dose in organ. Med Phys 2014;41:112105. Irazola L, Praena J, Fernandez B, Macias M, Terrón JA, Bedogni R, SánchezNieto B and Sánchez-Doblado F. Monitoring the stability of a thermal neutron detector using a moderated neutron beam from a Tandem Peletron. Appl Radiat Isot 2016;107:330-334. Irazola L, Terrón JA, Bedogni R, Pola A, Lorenzoli M, Sánchez-Nieto B, Gómez F and Sánchez-Doblado F. Improving the neutron-to-photon discrimination capability of detectors used for neutron dosimetry in high energy photon beam radiotherapy. Appl Radiat Isot, in press. Irazola L, Terrón JA, Bedogni R, Jiménez-Ortega E, Barbeiro AR, SánchezNieto B and Sánchez-Doblado F. Neutron measurements in radiotherapy: a method to correct neutron sensitive devices for parasitic photon response. Sent to Appl Radiat Isot. Irazola L, Terrón JA, Sánchez-Nieto B, Bedogni R and Sánchez-Doblado F. Neutron model upgrade for peripheral neutron dose assessment evaluated in 510 radiotherapy patients. Sent to Phys Med Biol. Irazola L, García-Hernández MT, Terrón JA, Ortiz-Seidel M, Velázquez S, Linares R, Sánchez-Nieto B, and Sánchez-Doblado F. SBRT, FFF and 10 MV irradiation techniques are associated to the lowest second cancer risk. To be sent to Radiat Oncol J. CMA National Instruments. Best application, 1st national prize Irazola L, Lorenzoli M, Bedogni R, Pola A, Gentile A, Terrón JA, and Sánchez-Doblado F. Helping to reduce patient risk of second cancer. CMA 2014, National Instruments (http://sine.ni.com/cs/app/doc/p/id/cs-15867). List of publications as a consequence of the thesis iv Full papers as collaborator Sánchez-Nieto B, El-far R, Irazola L, Expósito MR, Lagares JI, Mateo JC, Terrón JA and Sánchez-Doblado F. Analytical model for photon peripheral dose estimation in radiotherapy treatments. Biomed Phys Eng Express 2015;1:045205. Romero-Expósito M, Sánchez-Nieto B, Irazola L and Sánchez-Doblado F. In regard to “Neutron contamination in radiotheraphy: estimation of second cancers based on measurements in 1377 patients”. Sent to Radiother Oncol. Sánchez-Nieto B, Romero-Expósito M, Terrón JA, Paiusco M, Cagni E, Ghetti C, Filice S, Irazola L, Gómez F, Domingo C and Sánchez-Doblado F. Risk assessment of second cancer indicence after intensity-modulated radiation therapy and volumetric modulated arc therapy versus conventional conformal techniques. Sent to Br J Radiol. Short publications as first author Irazola L, Sanchez-Doblado F, Sánchez-Nieto B, Expósito MR, Mazzotti G, Morelli M, Lorenzoli M, Bedogni R, Pola A and Terrón JA. Evaluation of peripheral neutron equivalent dose and second cancer risk in radiotherapy patients. Radiother Oncol 2014;111:708-709. Irazola L, Lorenzoli M, Terrón JA, Bedogni R, Pola A, Sánchez-Nieto B, Romero-Expósito M and Sánchez-Doblado F. Neutron model upgrade for radiotherapy patients monitoring using a new online detector. Med Phys 2014;41:280. Irazola L, Terrón JA, Bedogni R, Lorenzoli M, Pola A, Sánchez-Nieto B and Sánchez-Doblado F. Signal photon component of a new thermal neutron detector TNRD in radiotherapy environments. Radiother Oncol 2015;115(1):S870(EP-1589). Irazola L, Terrón JA, Bedogni R, Lorenzoli M, Pola A, Sánchez-Nieto B and Sánchez-Doblado F. TNRD neutron detector signals for different gantry angles in 6 and 15 MV. Radiother Oncol 2015;115(1):S761 (EP-1410). Irazola L, Terrón JA, Sánchez-Nieto B, Bedogni R, Gómez F and SánchezDoblado F. Effects of cable extension and photon irradiation on TNRD neutron detector in radiotherapy. IFMBE Proceedings 2015;51:645-648. List of publications as a consequence of the thesis v Irazola L, Ortiz-Seidel M, García-Hernández MT, Terrón JA, Sánchez-Nieto B and Sánchez-Doblado F. Peripheral neutron dose estimation: comparison between experimental measurements and TPS estimation. IFMBE Proceedings 2015;51:397400. Irazola L, Terrón JA, Sánchez-Nieto B, Ortiz-Seidel M and Sánchez-Doblado F. Photon and Neutron Peripheral Dose Ratio for Low (6 MV) and High (15 MV) Energy for Treatment Selection. Med Phys 2015;42(6):3476. Irazola L, Domingo C, Romero-Expósito M, García-Fuste M, Terrón JA, Sánchez-Nieto B, Bedogni R and Sánchez-Doblado F. Estimation of Neutron Ambient Dose Equivalents for Radioprotection Exposed Workers in Radiotherapy Facilities Based On Characterization Patient Risk Estimation. Med Phys 2015;42(6):3417. Irazola L, Brualla L, Roselló J, Terrón JA, Sánchez-Nieto B, Bedogni R and Sánchez-Doblado F. Commissioning the neutron production of a Varian TrueBeam linac. Med Phys 2015;42(6):3376. Irazola L, Ortiz-Seidel M, Velázquez S, García-Hernández MT, Terrón JA, Sánchez-Nieto B, Romero-Expósito M, Roselló J and Sánchez-Doblado F. Comparison of peripheral doses associated to SBRT, VMAT, IMRT, FFF and 3D-CRT plans for lung cancer. ePoster at 35th ESTRO Congress 2016; EP-1613. https://www.postersessiononline.eu/pr/aula_poster.asp Irazola L. A new online detector for estimation of peripheral neutron equivalent dose in organ. Rev Fis Med 2015;16(3):60. Irazola L, Terrón JA, Sánchez-Nieto b, Romero-Expósito M and SánchezDoblado F. Peripheral neutron dose model verification for real IMRT cases. Poster at the 1st European Congress of Medical Physics (ECMP) 2016. Short publications as collaborator Terrón JA, Irazola L, Lorenzoli M, Bedogni R, Pola A, Introini MV, Bortot D, Gentile A, Esposito A, Sánchez-Nieto B, Expósito MR and Sánchez-Doblado F. Set-up of a new online digital detector for peripheral neutron equivalent dose estimation in radiotherapy patients. Radiother Oncol 2014;111:564. List of publications as a consequence of the thesis vi Sánchez-Doblado F, Irazola L, Lorenzoli M, Pola A, Bedogni R, Gentile A, Lagares, JI Muñiz JL, Sansaloni F, Introini MV, Bortot D, Sánchez-Nieto B, Expósito MR, and JA Terrón. Online neutron fluence measurements in phantom for second cancer risk estimation in radiotherapy. Radiother Oncol 2014;111:709-710. Sánchez-Nieto, El far R, Romero-Expósito M, Lagares J, Mateo JC, Terrón JA, Irazola L and Sánchez-Doblado F. Analytical model for Photon Peripheral Dose in Radiotherapy Treatments. Med Phys 2014;41:231. Sánchez-Nieto B, El-far R, Expósito MR, Lagares JI, Mateo JC, Terrón JA, Zeladas G, Irazola L and Sánchez-Doblado F. Peripheral dose assessment after IMRT and VMAT vs CFRT. Phys Medica: EJMP 2014;30:e33. Praena J, Irazola L, Fernández B, Terrón JA, Bedogni R, Lorenzoli M, Pola A, Sánchez-Nieto and Sánchez-Doblado F. Proposal of thermal neutron detector stability for peripheral dose estimation in clinic at a novel neutron facility. Radiother Oncol 2015;115(1):S735. Terrón JA, Irazola L, Morilla Y, Muñiz G, Bedogni R, Lorenzoli M, Pola A, Sánchez-Nieto B and Sánchez-Doblado F. Photon energy response of TNRD neutron detector in a 60Co irradiator and a 6 MV clinac. Radiother Oncol 2015;115(1):S757-S758. García Hernández MT, Ortiz M, Irazola L, Terrón JA, Romero-Expósito MR, Sánchez-Nieto B and Sánchez-Doblado F. Neutron peripheral dose estimation: treatment planning system implementation. Radiother Oncol 2015;115(1):S442. Sánchez-Nieto B, El-far R, Castrillón M, Irazola L, Terrón JA and SánchezDoblado F. Validation of a photon peripheral dose model for IMRT treatments. Radiother Oncol 2015;115(1):S541. Sánchez-Nieto B, Irazola L, Romero-Expósito M, Terrón JA and SánchezDoblado F. Validation of a peripheral photon dose model for clinical use: a prostate IMRT irradiation of the Alderson phantom. 35th ESTRO Congress 2016; PO-808. https://www.postersessiononline.eu/pr/aula_poster.asp Oral presentations Irazola L, Terrón JA, Bedogni R, Lorenzoli M, Sánchez-Nieto B and SánchezDoblado F. Gantry angle influence in the neutron commissioning of a linac. Oral Presentation at Congreso conjunto SEFM-SEPR 2015; Valencia. List of publications as a consequence of the thesis vii Sánchez-Doblado F, Terrón JA, Irazola L, Sánchez-Nieto B. Peripheral Neuton and Photon Doses. Oral presentation at IUPESM 2015; Toronto, Canada. Sánchez Nieto B, El Far R, Irazola L, Terrón JA, Sánchez-Doblado F. Peripheral photon dose in organs. Oral presentation at IUPESM 2015; Toronto, Canada. List of publications as a consequence of the thesis viii I. Preface 1. Radiation-induced secondary cancer 1 I. Preface 1. Radiation-induced secondary cancer Cancer represents one of the most important public health problems in developed societies. It has been estimated that about one in two men and women born today will be diagnosed with some type of cancer in their lifetimes (Newhauser et al.,2016; SEER website), being the leading cause of death among adults (23% of total deaths in 2012, Siegel et al.,2016). Close to 14 million of new cases were diagnosed in the world in 2012, and it is predicted that by 2030, this value will rise by 75% reaching nearly 24.6 million cases (Ferlay et al.,2015). The technological revolution that has taken place in the medical imaging field and diagnosis procedures, have allowed a better and earlier detection of cancer pathologies. In addition, the great improvements achieved in treatment techniques have led to an increased number of long-term survival patients (Coleman et al.,2011), as shown in Figure 1. For instance, the 5 year survival rate of cancer in the USA has grown to almost 68% and 83% in adults and children, respectively (Newhauser et al.,2016). Nowadays, there are three main weapons used to fight cancer disease, namely surgery, chemotherapy and radiotherapy. Although these treatment modalities are generally used in combination, almost two-thirds of cancer patients receive some form of radiation therapy during their treatment (Delaney et al.,2005; Jameson et al.,2013). In particular, considering the 10 most frequent cancer pathologies (prostate, female breast, lung, colorectal, bladder, non-Hodgkin lymphoma, skin melanoma, kidney, ovarian, and uterine), it is estimated that at least 50.3% of them are treated with radiotherapy as a part or the totality of the treatment (either X-ray based or with charged particles, Siegel et al.,2016; Delaney et al.,2005). I. Preface 1. Radiation-induced secondary cancer 2 Figure 1. SEER (Surveillance, Epidemiology and End Results program) database USA survival figures after cancer. Figure extracted from AAPM Congress 2015, Anaheim (http://amos3.aapm.org/abstracts/pdf/99-27178-365478-110196.pdf). Radiotherapy treatments are based on the use of ionizing radiation beams to destroy cancerous cells, obstructing their reproduction capability. The goal of this treatment modality is to deliver a high dose to the tumor (defined as Planned Target Volume, PTV) while surrounding organs (named Organs At Risk, OAR) and healthy tissues remain under certain established dose levels. One of the most common radiotherapy techniques, named external radiotherapy (cited hereafter as RT), consists in the use of an external radiation source, located outside the patient. This modality remains as the most used one, with an estimated number of 7 million cases for which radiotherapy was indicated in 2012 and a prediction of 12 million in 2030 (Atun et al.,2015). However this dual goal in an ideal scenario, where the PTV receives the 100% of the prescribed dose while the rest of the tissues remain at zero, is not feasible. This is mainly due to the unavoidable dose deposition at the beam entrance, lateral scattering and production of secondary uncharged particles (coming I. Preface 1. Radiation-induced secondary cancer 3 from the interaction of the primary beam with different materials along its path). The great benefits achieved with new RT techniques have been translated into higher healing rates, implying an extension of patient survival. This fact has led to a growing concern regarding second cancer incidence among long-term survivors, as they are at increased risk to develop treatment-induced side effects (Murray et al.,2015a;b; Hall et al.,2003; Xu et al.,2008; Berdnarz et al.,2010; Milecki et al.,2009; Trott et al.,2009a; Dörr et al.,2002a;b; Ruben et al.,2008; Brenner et al.,2000;2006; Tubiana et al.,2009;Murray et al.,2014; Hall et al.,2004; Chargari et al.,2013). These kind of radio-nduced malignancies have nothing to do with methastatic ones, coming from primary tumours. In general, Second Malignant Neoplasms (SMNs), account for around 17-19% of all cancers (Newhauser et al.,2016). As cancer patients have become younger, having thus a longer life expectancy after RT treatments, radiation effects are now more evident than in the past (Smith et al.,2010; Travis et al.,2003; Chao et al.,2016; Hung et al.,2016; Sharma et al.,2008). This increased risk of SMNs in healthy tissues as a consequence of RT, may appear usually after a period greater than a decade and, in around 10-20% of cases, 30 years after treatment (Tubiana et al.,2009; Newhauser et al.,2011; Kumar et al.,2012). This long-term side effect, although unusual, is a feared later complication of this “double-edged sword” that RT represents. Second cancer risks seem to be higher for tissues receiving low doses (≤6 Gy) and highly dependent on the dose deposited outside the PTV (Diallo et al.,2009; Dörr et al.,2002b). Treatment selection is usually based on several factors (i.e. life expectancy, clinical stage and associated toxicities; Robles et al.,2012; Bentzen et al.,2010; Shuryak et al.,2009), whereas these non-desired doses aren’t frequently considered in detriment of a benefit-risk ratio. Although second cancer risks attributable to radiotherapy are small (≈8%), bearing in mind the large number I. Preface 2. Peripheral dose: 1. Photon peripheral dose 10 result 5 times those of brachytherapy (regarding peripheral organs), around 100 times those of conventional radiodiagnosis images and on the order of radioimmunotherapy ones (concerning gamma doses; Xu et al.,2008). 2.1 Photon peripheral dose Photon doses delivered outside geometrical field limits and surrounding tissues are referred to as Peripheral Photon Doses (PPD). In clinical photon beams, these doses are produced by photons originated from: (i) linac head leakage, (ii) linac head scatter and (iii) scatter radiation inside the patient (Chofor et al.,2012). Although this component is the most studied one, as a consequence of the wellestablished photon dosimetry procedures, its estimation in peripheral organs is complex. Despite the fact that there are some algorithms and software available for the calculation of photon absorbed doses (Taddei et al.,2013; Jagetic et al.,2013; Van der Giessen et al.,2001; Hauri et al.,2016), to our knowledge there was no general model that could be readily applied to any RT treatment. The mentioned methods seem to be too narrowly focused, either in terms of available energies or treatment field configurations. What is more, although there are other studies containing baseline data for specific treatment techniques, they seem to be not generalizable to other circumstances. As an appropriate study of PD and second cancer risks may consider both contributions, our group was also highly interested in the evaluation of these PPD. Thus, along the duration of this work, close collaborations have been carried out with Dr. Beatriz Sánchez-Nieto for the generation of a clinically useful simple PPD model, applicable to any isocentric treatment and usable beyond 10 cm far from the field edge (Sánchez-Nieto et al.,2015a;b). This model accounts for linac head leakage, patient and linac head scatter, as well as activation, if any. I. Preface 2. Peripheral dose: 2. Neutron peripheral dose 11 2.2 Neutron peripheral dose The use of high photon energies is generally preferred to improve treatment effectiveness and decrease skin dose in deeply settled tumors. For this purpose, linacs up to 25 MeV (primary electron beams) were originally manufactured. However, as these photon beams are obtained by terms of electron bremsstrahlung in a metallic target with a high atomic number (e.g., tungsten, gold) there is an unavoidable neutron production. These primary neutrons are approximately isotropic and can penetrate the shielding (usually tungsten or lead) in all directions. Additionally, secondary neutrons can be produced in photonuclear reactions with the high Z materials that compose linac head for photon shielding and collimation purposes (Figure 4). Electronuclear (e,e’n) and photonuclear (γ,n) reactions can also take place as a consequence of the interaction of high-energy photons (greater than photonuclear reaction threshold of the present materials) with Fe and Cu (NCRP-79). The minimum energy required to remove one neutron from a nucleus lies between 6-16 MeV (6-13 MeV for most stable nuclei, those heavier than carbon), for the majority of constituents of RT bunkers. As a consequence, a wide variety of short-life radionuclides are generated via photon and neutron activations. In order to minimize this contribution, generally materials present in RT facilities are chosen to have a negligible neutron production for megavoltage photon modalities below 10 MV (Uselmann et al.,2015). Although neutrons can be also produced by other reactions as (γ,2n) and (γ,pn), they have smaller cross sections and are thus considered to be less relevant (Sánchez-Doblado et al.,1989). Different works have challenged the estimation of Peripheral Neutron Doses (PND) from a clinical point of view. Although their initial conclusion showed a minimal importance for exposed professionals and patients, later publications I. Preface 2. Peripheral dose: 2. Neutron peripheral dose 12 Figure 4. Sketch of head components of a Siemens Primus linac geometry. The different components responsible for photo-neutron production in high energy (>10 MV) are plotted: lead (black), tungsten (grey) and steel (light grey). Figure extracted from (Pena et al.,2005). presented a more precise scenario of neutron energy and fluence distribution in treatment rooms, related to linac manufacturer, energy and bunker size (Kry et al.,2009; Howell et al.,2005;2009; Followill et al.,2003). However, as previously mentioned, this is hardly ever considered in clinical routine, as there is no commercial TPS that includes information regarding neutron contribution. This fact can be justified by the high photon doses delivered at the studied locations, compared to the peripheral neutron ones. However, PND may become equally important than PPD for some peripheral tissues, due to the higher neutron Relative Biological Effectiveness (RBE, Ottolenghi et al.,2015). As a consequence, modern RT is mainly oriented to low energy I. Preface 2. Peripheral dose: 2. Neutron peripheral dose 13 treatments, sometimes to the detriment of conformity and using longer beamon times. This is justified by the pretext of avoiding damage to biological tissues and preventing from latter complications due to neutrons. 2.2.1 Evaluation of neutron dose: detectors Neutron dose measurement inside radiotherapy rooms is a mandatory task that has to be performed accomplishing some requirements (Thomas et al.,1973): (i) Measurement of neutron fluence (n·cm-2) (ii) Measurement/simulation of neutron spectrum as a function of energy inside the treatment room (n·cm-2·MeV-1·s-1) (iii) Measurement of total dose equivalent (iv) Study of neutron detector response to other radiations that may be present in RT mixed fields Neutron energy spectra present in RT environments can be divided into three regions: thermal (< 0.4 eV), epithermal (0.4 eV < E < 0.1 MeV) and fast (> 0.1 MeV). However neutron dosimetry is a very complex task and there is no single detector commercially available that can perform all the above detailed measurements. Consequently, detector choice must be done regarding the specific purpose of the experiment. As neutrons are uncharged particles, these devices are based in the detection of particles resulting from their interaction with atomic nuclei. The most commonly used neutron devices can be divided in two main groups, some of the most typical ones are shown in Figure 5 as an example (NCRP-79). I. Preface 2. Peripheral dose: 2. Neutron peripheral dose 14 (a) Passive detectors These kinds of detectors are the most used ones due to their good response to the pulsed nature of RT radiation. However their complex and delayed reading procedures make their use non-ideal for research or clinical purposes, where measurement repetitivity and online behavior is an important aspect. - Activation detectors. These devices are based on the use of specific materials that become radioactive when exposed to neutrons (Figure 5a). This behavior may result from capture reactions as inelastic scattering; (n,p) (n,α) or (n,2n) reactions; nuclear reactions or spallation reactions. Activation samples are typically used to characterize energy spectrum and neutron field intensity. - Etched Track Detectors. These types of detectors are composed of a particular kind of solids that present microscopic radiation damage tracks when irradiated with heavy particles. These tracks can be made visible under a microscope in a variety of dielectric materials (Figure 5b). (b) Active detectors These detectors present an online behavior, highly desirable for research purposes. Nevertheless, several problems related to saturation, response dependencies and detector size can be found in the existing devices. - Ionization Detectors. Ionization chambers, and proportional and Geiger-Müller counters, are included here (Figure 5c). They provide large electronic signals and good discrimination against a relatively small photon background, being usually employed for ambient measurements due to their large size. - Diodes. Neutron-induced nuclear reactions can produce measureable changes in the electrical properties of various semiconductors (Figure 5d). The most common radiation damage is the displacement of lattice atoms, as in the case of diode detectors. I. Preface 2. Peripheral dose: 2. Neutron peripheral dose 15 Figure 5. Passive: (a) Activation foils, (b) track-edge and Active: (c) proportional counters and (d) diode detectors. Passive detectors require considerable processing and analyzing times while active counters show signal saturation and pile-up due to the pulsed nature of linac radiation. Therefore, methods traditionally used for neutron evaluation have been the first ones, as recommended by the American Association of Physicists in Medicine (AAPM) and the National Council on Radiation Protection & measurements (NCRP, Nath et al.,1986). However, our group succeeded in the introduction of a new active detector in RT environments for peripheral neutron dose estimation purposes (Gómez et al.,2010a; further details can be found on section 2.4.2). (a) (b) (c) (d) I. Preface 2. Peripheral dose: 2. Neutron peripheral dose 16 2.2.2 Evaluation of neutron dose: calibration sources As for any other detectors, neutron devices have to be calibrated (and tested in terms of stability) against reference neutron sources to ensure appropriate estimations. A wide variety can be employed for this purpose (NCRP-38;151): - Natural radioactive neutron sources type (α,n) or (γ,n). In the first case, the alpha emitter is mixed with the target material, being the most common ones 210Po, 239Pt and 241Am, usually combined with Be. The latter are generally surrounded by the material in which neutron reaction takes place. - Spontaneous fission neutron sources. They present small mass and a fission-like spectrum with relative low gamma-ray yield. Special caution must be taken when manipulating, as many alpha rays are produced per neutron. A nuclide of special interest is 252Cf. - Constant voltage accelerators. Used to produce high voltage proton beams by mechanical transportation of charge using a belt (Van der Graaff accelerator) or other alternative systems that imping into a specific target to produce neutrons. Large number of neutrons (>1022 s-1) can be achieved depending on incident beam and target nature. In some cases, like PTB laboratory (PhysikalischTechnische Bundesanstalt), monoenergetic beams can be achieved in the range of 24 KeV-19 MeV. - High-frequency positive ion accelerators (including cyclotron, synchrocyclotron, proton synchrotron and heavy ion linear accelerators). These pulsed beams can produce a neutron spectra spread over a wide range of energies. - High-frequency electron accelerators (including circular betatron and synchrotron and linear accelerators). Although primary radiation are X-rays, these accelerators can produce large numbers of neutrons depending on target nature, by terms of (γ,n) reactions. I. Preface 2. Peripheral dose: 2. Neutron peripheral dose 17 - Nuclear reactors. Here neutron production occurs as a result of the fission process, achieved by a definite arrangement of fuel (uranium or plutonium), moderator (light or heavy water, beryllium or graphite) and neutron absorbers. Neutron energies range from thermal to over 15 MeV, with many gamma rays present. Some studies have already pointed out the importance of primary and secondary reference neutron fields for calibration of these instruments as necessary for advances in research methodologies (Newhauser et al.,2016). An additional problem concerning the use of neutron detectors in radiotherapy environments is the great change of the neutron spectra with increasing depth in tissue. This can generate misleading in readings of some detectors when used for different locations for ‘in-phantom’ measurements. Thus, the ideal choice for neutron measurements in clinical environments would be a combined use of different devices. Besides, the variable attenuation and scattering that neutrons undergo when trespassing human body implies the need of MC simulations to fully determine neutron fluences at any point. MCNPX code is typically used to simulate neutron spectra inside RT rooms and patients. Calculation of this neutron spectrum may need information related to bunker and phantom geometries as well as linac composition, in order to obtain the most realistic photoneutron spectrum possible. Moreover, neutron photoproduction strongly depends on nominal energy of initial electrons, and a whole commissioning process, with accurate information about linac head composition is also desirable (González-Soto et al.,2011). Thus, in order to provide adequate PND estimations, data should be based on a combination of both experimental measurements and simulations in air and phantom locations. I. Preface 2. Peripheral dose: 3. Equivalent neutron doses and risk evaluation 18 2.3 Equivalent neutron doses and risk evaluation As previously stated, the RBE factor highly depends on radiation type (e.g., photons, neutrons, electrons) as well as particle energy, dose and biological endpoint. Peripheral dose estimations previously detailed represent a former step for the evaluation of Second Cancer Probability (SCP) related to radiotherapy treatments. Photoneutron spectrum varies along the patient, presenting thermal, epithermal and fast neutron fluence components; the latter especially important for superficial points and decreasing for deeper positions, due to neutron thermalization. In consequence, for an adequate organ neutron dose estimation, experimental measurements and MC simulations should be performed for different locations and depths in anthrophomorphic phantoms. However, as previously explained, there is no single detector able to measure total neutron fluences and they are usually estimated in terms of the thermal component, which is the simplest one to measure. For PND estimations in radiotherapy environments, dose equivalent at a point and organ-equivalent dose are the most significant magnitudes (Romero-Expósito et al.,2016). The first one can be defined as follows: 𝐻𝑇=𝐷·𝑄=∫ ∫ 𝑄(𝐿𝐸𝑇)·𝐷𝐿·𝑑𝐿·𝑑𝑚 𝑚 𝐿𝐸𝑇𝑚 (eq. 1) where DL represents absorbed dose, Q is the quality factor for the specific radiation and integrals range over unrestricted Linear Energy Transfer (LET) and mass of the studied organ (m). Although Q factor is difficult to evaluate due to its dependence on particle energy, recent publications (Romero-Expósito et al.,2016) demonstrated that dose equivalents calculated using Q and the much simpler radiation weighting factor wR (introduced by ICRP-103) are compatible. Thus eq. 1 can be rewritten as: I. Preface 2. Peripheral dose: 3. Equivalent neutron doses and risk evaluation 19 𝐻𝑇=∑𝑤𝑅·𝐷𝑇,𝑅 𝑇 (eq. 2) According to ICRP definition, Equivalent Dose (H [Sv]) is a protection quantity calculated as the averaged absorbed dose (D [Gy]) in an organ or tissue T multiplied by radiation weighting factor wR (ICRP-103). For low-dose levels, this factor is used as a conservative and simplified measure of RBE (Figure 6), taking into account different particle nature, concerning radiobiological effect (wR(photons/e)=1, wR(neutrons)=2.5-20 depending on their energy). On the other hand, Effective Dose (E [Sv]) corresponds to the mean dose in organs or tissues of the whole human body, weighted by their radiosensitivity. This quantity is usually employed to indicate whole-body stochastic risk. Figure 6. Different radiation weighting factors and dependence with neutron energy. Extracted from: http://www.eurados.org/~/media/Files/Eurados/events/7thwinterschool/4_Menzel%20Hans_Cu rrent%20Approach%20%20to%20Radiation%20Quality%20Specification%20in%20Radiatio n%20Protection.pdf?la=en. I. Preface 2. Peripheral dose: 4. Peripheral neutron dose models 26 2.4.3 Anthrophomorphic phantom measurements. NORMA phantom An anthrophomorphical adult phantom was designed to mimic neutron interaction with human body. In order to correctly simulate this behavior of human tissue, several materials were previously studied (Domingo et al.,2011). Finally, polyethylene was chosen as a surrogate of human tissue and lowdensity wood was used to simulate lungs (Figure 11). For organ dose estimation, sixteen accommodations for passive detectors (plastic and thermoluminiscent, TLD) were chosen as representative for organ location in the human body (Sánchez-Doblado et al.,2012). Correspondence among the 16 NORMA measuring points and patient organs for is shown in the Table of Figure 11. These correspondences were done by comparing NORMA anthrophomorphic phantom to the reference mathemathical one, named Cristy. 2.4.4 Neutron equivalent doe in organ and second cancer risk estimation Neutron equivalent doses for each point in NORMA phantom were evaluated, from eq.2, considering kerma approximation 2 and using previously mentioned convolution procedure, described in the following equation: 𝐻𝑇=𝛷𝑡ℎ∫𝑘(𝐸)·𝜑(𝐸)·𝑤𝑅(𝐸)𝑑𝐸 𝐸 (eq. 4) 2 Absorbed dose can be approximated by factor kerma as follows: D=k·Φ. Equivalent dose can be then calculated by: HT=wR·D introducing neutron radiation weighting factor (wR; ICRP-103). Differences between dose equivalent and equivalent doses are much smaller than neutron measurement uncertainties making this approximation as acceptable for this purpose (Romero et al.,2016) I. Preface 2. Peripheral dose: 4. Peripheral neutron dose models 27 Figure 11. Scheme of the NORMA phantom composition and point location. Table in the right correlates these points with evaluated patient organs. Figure modified from Sánchez-Doblado et al.2012; Expósito et al.,2013. where: - Φth: thermal neutron fluences measured by the passive detectors located inside the phantom - k(E): kerma factor for ICRU tissue (Siebert et al.,1995) - 𝜑(E): MC neutron fluence energy spectra, normalized to the thermal contribution - wR: energy-dependent radiation weighting factor from ICRP-103 I. Preface 2. Peripheral dose: 4. Peripheral neutron dose models 28 Two types of passive detectors (PADC and TLD) were used for the measurement of neutron fluences at the 16 i-points inside NORMA for the two j-studied treatments (Sánchez-Doblado et al.,2012). Due to their different sensitivity to thermal, epithermal and fast components, a combination of both readings was used to correctly rescale total neutron fluence Φ from MC simulations, depending on treatment location and specific point. The methodology established by this group for peripheral neutron dose calculation, was designed to directly estimate SCP to 13 organs from the total number of high-energy MU received by the patient during the whole treatment, differentiating between two different treatment locations (Sánchez-Doblado et al.,2012; Expósito et al.,2013). A linear relation depending only on facility characterization, in terms of neutron production, was obtained. Thus, a simple commissioning for every facility is needed. This procedure consists in 3 single measurements at the reference location with a SRAM-based detector (Romero et al.,2015). Thus SCP, in cases per 1000, for an organ -k using model –j, can be calculated as follows (Sánchez-Doblado et al.,2012): [𝑆𝐶𝑃]𝑘,𝑗(𝑐𝑎𝑠𝑒𝑠 1000)=𝑐(𝑒𝑣𝑒𝑛𝑡𝑠∗ 1000 )·𝑁(𝑀𝑈)·𝑠·[[𝑔]𝑘,𝑗(𝜇𝑆𝑣 𝑒𝑣𝑒𝑛𝑡)·[𝜆]𝑝,𝑘(% 𝑝𝑒𝑟 𝑆𝑣) 10 ] (eq. 5) where: - c represents the characterization factor, obtained for each facility by following the procedure detailed in Romero-Expósito et al.,2015. This factor has two dependencies: one in terms of detector used, which has to be referred to events of the reference device (FC) and the other related to bunker surface (FA), taking into account the effect of room size on thermal neutrons distribution. - N is the total number of high energy MU delivered in one session of the treatment. I. Preface 2. Peripheral dose: 4. Peripheral neutron dose models 29 - s is the total number of sessions that contain high energy along the whole treatment. - gk,j are the factors that correlate neutron equivalent dose in organ -k (μSv) to events of the digital reference detector for each treatment –j (Table I). - 𝜆𝑝,𝑘 risk factors for organ -k tabulated by -p protocol (BEIR VII or ICRP103), that correlate neutron doses to second cancer risk probabilities. - factor 10 was introduced to express the result in cases per 1000. According to this expression, the previously mentioned locations were considered to encompass the most common RT treatments. Then, gk,j factors were estimated from the combination of MCNPX simulations and passive detectors readings, giving values presented in Table I. Table I. Neutron equivalent dose in organ (μSv) per event at the SRAM-based detector, corrected by the area factor at digital device (gk,j), for the two general treatments. Sex is considered when selecting affected organs. Extracted from (Sánchez-Doblado et al.2012; Expósito et al.,2013). Organ Head Treatment Abdomen treatment Thyroid 0.71 ± 0.21 0.37 ± 0.11 Esophagus 0.45 ± 0.13 0.65 ± 0.19 Lung 1.32 ± 0.40 2.22 ± 0.67 Breast 0.66 ± 0.20 1.62 ± 0.49 Stomach 0.22 ± 0.07 0.55 ± 0.16 Liver 0.23 ± 0.07 0.69 ± 0.21 Colon 0.05 ± 0.01 0.21 ± 0.06 Urinary bladder 0.27 ± 0.08 1.11 ± 0.33 Ovary 0.05 ± 0.01 0.21 ± 0.06 Skin 2.64 ± 0.79 2.64 ± 0.79 Bone surface 0.50 ± 0.15 0.96 ± 0.29 Red marrow 0.57 ± 0.17 1.90 ± 0.57 Remainder 0.38 ± 0.11 0.41 ± 0.12 These models allow the estimation of SCP (cases per 1000) for 13 organs for two different treatment locations. However, the appearance of new online active I. Preface 2. Peripheral dose: 4. Peripheral neutron dose models 30 miniaturized thermal neutron detectors, open the door to further investigations in order to improve and generalize these models. The advantage of the new detectors chosen, named TNRD (Thermal Neutron Rate Detector), was the possibility of performing online measurements at external and ‘in-phantom’ locations, directly in terms of thermal neutron fluences with the same device (Bedogni et al.,2014). This would allow the generalization of the established methodology to be expressed in terms of thermal neutron fluence, without the need of referring data to the reference detector (in terms of SEU, Gómez et al.,2010b; Domingo et al.,2010a; Sánchez-Doblado et al.,2012). In addition, the active and miniaturized behavior would allow the verification and optimization of these models, as well as the introduction of patient anatomy and treatment technique influences (different locations and available techniques). II. Hypothesis and objectives 31 II. Hypothesis and objectives The acquired new miniaturized active thermal neutron detectors (TNRD), initially designed for nuclear purposes, can be used for patient dosimetry in radiotherapy. The use of these devices for peripheral neutron dose estimations would be useful not only for online patient measurements following the previously established methodology, but also for ‘in-phantom’ measurements. This would allow the improvement of the previously existing models, consisting on Head & Neck and abdomen locations for adult phantom, to more patient specific ones, including patient size. This methodology can be also generalized to be expressed in terms of thermal neutron fluences, thanks to its dual use. This would represent a previous step to model implementation in Treatment Planning Systems for risk evaluation in clinical routine. Peripheral neutron and photon doses could be then considered as an objective decision tool for treatment selection strategy. Thus, the main objectives of this work are: 1. Validation and characterization of the new TNRD thermal neutron detectors with the previously existing ones for their use in radiotherapy environments for both patient and ‘in-phantom’ measurements. 2. Guarantee the validity of the existing models by terms of the new existing detectors. 3. Adapting detectors setup to clinical environments considering electronics capability, as well as improving photon rejection for radiotherapy backgrounds. 4. Enhancement of the existing models ensuring the reproducibility of ‘inphantom’ measurements. II. Hypothesis and objectives 32 5. Generalization of the currently existing models for peripheral neutron dose estimations, to express them directly in terms of thermal neutron fluence. Introduce patient dimensions by measuring different phantom sizes, performing patient measurements and comparing estimations for specific real cases. 6. Clinical implication: evaluation of peripheral doses (neutron and photon) for different treatment locations, techniques, energies and linacs. Establishing a model that can be applied in the clinic for peripheral dose assessment and second cancer risk estimations. III. Thesis core 1. Introduction: (A) Characterization of a new active thermal neutron detector 33 III. Thesis core This section describes the body of this work, firstly introducing the necessity and purpose of each experiment. Then, a brief description of the methodology and measurements carried out will be presented. Finally, a selection of the main related publications concerning each topic will be added to illustrate accomplished results and conclusions. A total of six full papers have been added within the body of the text (framed and highlighted in light beige) in this chapter, being published, accepted or sent to journals in the first tercil. In addition, seven short publications containing partial results, also published in indexed journals and two in the World Congress on Medical Physics & Biomedical Engeneering as full paper, have been included in the Appendix. 1. Introduction (A) Characterization of a new active thermal neutron detector The newly acquired Thermal Neutron Rate Detector (TNRD) is a miniaturized active thermal neutron detector initially designed for nuclear purposes. Therefore former to its introduction for peripheral dose measurements, the acquired prototypes had to be tested and adapted for clinical routine. For that: (i) Detectors were irradiated under several conditions in order to verify their adequate response in RT environments. TNRD readings were correlated to those of the SRAM-based reference detector, in order to establish a calibration factor. In addition, as pulsed neutron signals have to be measured under an intense photon background, linear response and reproducibility were studied under these specific conditions (Terrón, Irazola et al.,2014, Appendix A.1). III. Thesis core 1. Introduction: (B) Limitations and solutions of TNRD detectors in radiotherapy environments 34 (ii) TNRD ‘external’ thermal neutron measurements, were compared to those of the reference detector by using the established methodology for the online assessment of second cancer risk in radiotherapy patients (Irazola et al.,2014a, Appendix A.2). (iii) In order to use these devices for the enhancement of the existing models, consisting in the correlation among readings of a reference detector (located in the room) and neutron dose estimations in several phantom points (Sánchez-Doblado et al.,2012; Expósito et al.,2013), ‘inphantom’ measurements had to be performed with TNRD devices. Preliminary results were compared to those of TLD devices, previously used for this purpose (Sánchez-Doblado F, Irazola L et al.,2014, Appendix A.3). (iv) Thermal neutron fluences were evaluated with TLD and TNRD detectors in all the phantom points for the two general treatments (H&N and abdomen), as well as a real IMRT one. Once both ‘external’ and ‘in-phantom’ uses of TNRD detectors were validated against reference detectors, we were able to ensure the use of these new devices for the final purpose of improving the existing peripheral neutron dose models. These global results, regarding TNRD use in radiotherapy environments, were published in Medical Physics journal (Irazola et al.,2014b, Section III.2.A). (B) Limitations and solutions of TNRD detectors in radiotherapy environments The prolonged use of TNRD devices for ‘external’ and ‘in-phantom’ measurements in clinical routine, showed several incongruences in detector response over time: III. Thesis core 1. Introduction: (B) Limitations and solutions of TNRD detectors in radiotherapy environments 35 (i) A different loss in sensitivity to thermal neutron fluences was noticed for each one of the six available TNRD detectors after 9 months of continuous use. Meaning a possible different detector ageing in time, we decided to recalibrate them in a new thermal neutron source available at INFN, Frascati (Bedogni et al.,2016). In addition, we thought that having the option of periodically verify their stability under an external neutron source would be highly desirable. The neutron beam available at Centro Nacional de Aceleradores (CNA, Sevilla), seemed to be a good option, considering the proximity of this facility and its relation to the University of Seville. Several measurement tests and MCNPX simulations were performed in order to achieve an appropriate thermal neutron beam and detector setup. The goal was to obtain a stable beam that provided thermal neutron fluences in the order of those from calibration conditions at detector location, avoiding photon presence as much as possible. Preliminary results were published as a short work in Radiother Oncol (Praena et al.,2015). Final configuration concerning not only the validation of this source for the study of TNRD stability verification, but also bringing to light the possibilities available at the Pelletron Tandem accelerator of CNA, was published as a full article in Applied Radiation and Isotopes (Irazola et al.,2016a, Section III.2.B.1). (ii) Uncertainties introduced by the possible loss of signal, as a consequence of the necessary cable extension and/or cable irradiation throughout ‘in-phantom’ measurements may also represent a minor problem. Initial TNRD cable length showed problems when trying to reach different NORMA phantom points at the same time. Thus, we decided to elongate them and improve device electronic to make it III. Thesis core 1. Introduction: (D) Peripheral dose in clinical cases 42 III. Thesis core 2. Publications: (A) A new online detector for estimation of perihpheral neutron equivalent dose in organ 43 2. Publications (A) A new online detector for estimation of peripheral neutron equivalent dose in organ Irazola et al., Med Phys 2014;41:112105 Abstract Purpose: Peripheral dose in radiotherapy treatments represents a potential source of secondary neoplasic processes. As in the last few years, there has been a fast-growing concern on neutron collateral effects, this work focuses on this component. A previous established methodology to estimate peripheral neutron equivalent doses relied on passive (TLD, CR39) neutron detectors exposed in-phantom, in parallel to an active [static random access memory (SRAMnd)] thermal neutron detector exposed ex-phantom. A newly miniaturized, quick, and reliable active thermal neutron detector (TNRD , Thermal Neutron Rate Detector) was validated for both procedures. This first miniaturized active system eliminates the long postprocessing, required for passive detectors, giving thermal neutron fluences in real time. Methods: To validate TNRD for the established methodology, intrinsic characteristics, characterization of 4 facilities [to correlate monitor value (MU) with risk], and a cohort of 200 real patients (for second cancer risk estimates) were evaluated and compared with the well-established SRAMnd device. III. Thesis core 2. Publications: (A) A new online detector for estimation of perihpheral neutron equivalent dose in organ 44 Finally, TNRD was compared to TLD pairs for 3 generic radiotherapy treatments through 16 strategic points inside an anthropomorphic phantom. Results: The performed tests indicate similar linear dependence with dose for both detectors, TNRD and SRAMnd, while a slightly better reproducibility has been obtained for TNRD (1.7% vs 2.2%). Risk estimates when delivering 1000 MU are in good agreement between both detectors (mean deviation of TNRD measurements with respect to the ones of SRAMnd is 0.07 cases per 1000, with differences always smaller than 0.08 cases per 1000). As far as the in-phantom measurements are concerned, a mean deviation smaller than 1.7% was obtained. Conclusions: The results obtained indicate that direct evaluation of equivalent dose estimation in organs, both in phantom and patients, is perfectly feasible with this new detector. This will open the door to an easy implementation of specific peripheral neutron dose models for any type of treatment and facility. 1. Introduction It is widely known the paradoxical fact that ionizing radiations, while being an effective therapeutic mean to fight cancer, also represent a potential source of new neoplasic processes. The study of risks associated to certain dose distributions is usually limited to the areas surrounding the tumor. However the rest of the body is similarly affected by the radio-induced toxicity with dose values that, although initially were considered insignificant, can be of the order of the ones that may affect any exposed professional. Although they have rarely been considered in the clinical routine of the institutions around the world, in the last few years, there has been a fast-growing concern about the peripheral radiation dose, outside the treatment volume, that patients receive additionally and unnecessarily. III. Thesis core 2. Publications: (A) A new online detector for estimation of perihpheral neutron equivalent dose in organ 45 Although in the case of photons (the main radiation contribution to the peripheral dose), there are well-known dosimetric mechanisms, there is not any comparable system to measure the neutron contribution (Xu et al., 2008). We have focused our studies on the latter, as it was the least analyzed because of its great complexity. During the last few years, some neutron detectors were developed and numerous experimental tests were done with medical linacs. The first results obtained with an active neutron detector, based on Static Random Access Memory (SRAMnd), indicated that the estimation of neutron dose in patients could be feasible (Gómez et al., 2010; Sánchez-Doblado et al.,2012). A first study on risk estimation for 1377 patients in 50 international facilities has been carried out for the 14 most frequent pathologies and 15 linac models (which virtually represent the totality of the existing machines; Expósito et al.,2012). The methodology presented in Sánchez-Doblado et al., 2012, is not restricted to a specific detector but applicable to any other. For example, to the new digital diode-based detector introduced by Guardiola et al.,2013. The aim of the present work is the validation of a novel miniaturized active digital detector named TNRD (Thermal Neutron Rate Detector), developed by Bedogni et al.2014, for its use in a clinical environment. The great advantage of this device is its dual use, both as online detector in clinical routine and for in-phantom measurements, due to its reduced size. 2. Material and Method 2.A. Digital detector TNRD The new thermal neutron detector, developed by the NESCOFI@BTF project (Scientific Commission V, INFN-LNF, Italy) is based on a commercial solidstate device sensitized to thermal neutrons through a customized physical- III. Thesis core 2. Publications: (A) A new online detector for estimation of perihpheral neutron equivalent dose in organ 46 chemical treatment. Details on design and composition of the detector can be found in Bedogni et al.,2014. With an active area of 1 cm2 and overall dimensions of approximately 1.5 x 1 x 0.4 cm3, it linearly responds in terms of thermal neutron fluence rate from 10 up to 106 cm-2·s-1. The TNRD signal is amplified in a low-voltage electronics module and sent to a PC-controlled programmable ADC. The control software was developed in LabView© (2010 National Instruments), by the Politecnico di Milano. Compared to the SRAMnd detector (Gómez et al.,2010), the TNRD can correctly measure lower intensity fields. The TNRD output is a DC voltage directly proportional to the thermal neutron fluence rate (Bedogni et al.,2014). The calibration factor is obtained for every TNRD by exposing it in a reference thermal field at INFN-Frascati. The detector-to-detector response variability is in the order of ±5% (1SD). An additional uncertainty contribution is needed to account for the dependence of the TNRD response from the energy and direction distribution of the incident field, so that a 10% overall uncertainty for the TNRD response was estimated. 2.B. Li thermoluminiscence dosimeters Standard 6LiF/7LiF pairs of TLD-600/TLD-700 (3 × 3 × 0.9 mm3 chips) dosimeters were used as an independent system to validate the thermal neutron fluences, obtained with TNRD detectors in different points inside a NORMA phantom (Sánchez-Doblado et al.,2012). Conventional neutron calibration was performed at Physikalisch-Technische Bundesanstalt in scattered neutron reference radiation fields, produced by a bare 252Cf and a D2O-moderated 252Cf neutron source and for gamma, using a 137Cs calibration source at a secondary standard dosimetry laboratory in CIEMAT (Centro de Investigaciones Energéticas Medioambientales y Tecnológicas). III. Thesis core 2. Publications: (A) A new online detector for estimation of perihpheral neutron equivalent dose in organ 47 On the basis of the results of glow curve analysis, thermal neutron fluences were obtained using the differences between TLD-600 and TLD-700 readings for each measured point in NORMA and the calibration factors as follows: 𝜙𝑖,𝑗 𝑇𝐿𝐷 =𝑓600/700 𝑛[𝑅600 𝑖,𝑗 −𝑘𝑅700 𝑖,𝑗 ] 𝑘=𝑓600 𝛾𝑓700 𝛾 ⁄ (eq. 1) Where the calibration factors used are 𝑓600/700 𝑛=465𝑛·𝑐𝑚−2 (12%, k=2) for neutrons and 𝑓600 𝛾=1.86𝑥10−4𝑚𝐺𝑦·𝑎𝑢−1 (8%, k=2) and 𝑓700 𝛾=1.99𝑥10−4𝑚𝐺𝑦· 𝑎𝑢−1 (10%, k=2) for gammas. The global uncertainty for these procedures has been estimated at 15%. In order to validate TNRD for its use in the methodology established by Sánchez-Doblado et al.,2012, it was compared to both: the active SRAMnd detector, used for ex-phantom measurements in clinical routine and passive ones (here TLD), which are used for in-phantom measurements, to generate risk models. 2.C. TNRD validation for clinical use First, the “reference factor”, defined as the quotient between the SRAMnd and the TNRD readings (unit: V-1·s-1), was studied by exposing the devices under the reference irradiation conditions in a linac: 10x10 cm2 field, 0° gantry angle and 1000 Monitor Units (MU), and detectors located in front of the gantry axis, close to the wall (Figure 1) as recommended in Jíménez-Ortega et al.,2011 (reference position for patient measurements). Thus, once this factor is obtained [with a better accuracy than in (Terrón et al., 2014) by increasing the number of measurements], measurements can be performed in any facility by applying the methodology established in Sánchez-Doblado et al.2012, to estimate patient neutron equivalent dose in organ. III. Thesis core 2. Publications: (A) A new online detector for estimation of perihpheral neutron equivalent dose in organ 48 Figure 1. (a) SRAMnd and TNRD detectors (the metallic box is the detector based on SRAM, while the black one is the new detector TNRD presented in this paper) and (b) Location of the detectors in the treatment room when NORMA phantom is used (Sánchez-Doblado et al.2012). The insert shows how TNRD detector is placed inside the phantom. (a) (b) III. Thesis core 2. Publications: (A) A new online detector for estimation of perihpheral neutron equivalent dose in organ 49 As previously mentioned (section 2.A), intrinsic characteristics of TNRD detector had already been studied under a pure neutron beam by the manufacturer. However this device is also sensitive to the photon component, which is highly present in radiotherapy environments, thus its validation under clinical conditions becomes essential. Reproducibility and linearity were studied in this environment and compared to the previously used SRAMnd device. A series of 18 measurements were performed with TNRD and SRAMnd detectors, under the above-mentioned reference irradiation conditions during six days (3 each day). Readings were corrected for linac stability by means of an ionization chamber placed inside the photon beam (this was not considered in previous measurements; Terrón et al., 2014). Linearity was studied for two of the detectors, in an extended range of typical monitor units (MU) of radiotherapy treatments (25 to 4000 MU) measured under the defined reference conditions. This study allows the knowledge of the minimum reliable measureable fluence. The next step was to compare their capabilities for facility characterization and patient measurements, for which this new detector would be useful from a clinical point of view. Characterization procedure (Expósito et al.,2013), was done for TNRD and SRAMnd in four different linacs (Terrón et al.,2014) (nominal energies ranging from 15 to 18 MV) in Hospital Universitario Virgen Macarena (HUVM) in Sevilla and Santa Maria delle Croci Hospital (SMCH) in Ravenna. By applying the reference factor to the readings of the TNRD, located in the reference position, neutron equivalent dose in organs was assessed as proposed in (Sánchez-Doblado et al.,2012). A cohort of 200 real patients from the 4 facilities, were evaluated in terms of second cancer risk with both detectors. Estimates were done by two alternative methods. The first one uses the models III. Thesis core 2. Publications: (A) A new online detector for estimation of perihpheral neutron equivalent dose in organ 50 obtained during the characterization of each facility, as a function of the delivered MU for each patient treatment. The second one makes the estimations by applying the methodology established in Sánchez-Doblado et al.,2012 to the direct readings from the detectors, placed in the treatment room during patient irradiation. 2.D. TNRD validation for in-phantom measurements The added new possibility of in-phantom measurements with an active detector for the first time has to be validated with the well-established methodology of classical passive detectors, sensitive to thermal neutrons [here TLDs; SánchezDoblado et al.2012]. Measurements were performed in a Siemens Primus linac at HUVM with detectors located inside the NORMA phantom [SánchezDoblado et al.2012 (Figure 1b)], to confirm their viability. Three treatments were considered: (i) Head and neck, eight beam incidences (at 0°, 45°, 90°, 135°, 180°, 225°, 270°, 315°), 10 × 10 cm2 field size, 1000 MU, isocenter in the middle of the head. (ii) Abdomen, same beam incidences, same field size and number of MU as in i, isocenter in the middle of abdomen, just in the border of the lower lungs. (iii) Prostate IMRT, seven beam incidences (at 0°, 52°, 95°, 156°, 204°, 265°, 308°), 420 MU, isocenter at the prostate level. 3. Results and discussion 3.A. TNRD validation for clinical use A mean reference factor of 53.55±1.54 events/V·s (1SD) has been obtained for the TNRD detector with respect to the SRAMnd, used as a reference. III. Thesis core 2. Publications: (A) A new online detector for estimation of perihpheral neutron equivalent dose in organ 51 The uncertainty in reproducibility of neutron fluence measurements, normalized to the mean value along the days, is lower than 1.7% (1SD) for TNRD, which is slightly better than the 2.2%(1SD) obtained for the SRAMnd detector. A well-fitting linear regression model (R2=0.9999) was obtained without saturation effects observed, for both detectors. Figure 2 shows the mean standard deviation (error bars for variations among both detectors used) for two of the six available TNRD detectors, as a function of thermal neutron fluence (logarithmic representation) in the typical radiotherapy treatment range. We can estimate that, considering an uncertainty over 5%, the minimum reliable thermal neutron fluence is about 0.19 x105/cm2, corresponding to approximately 5 MU in this reference conditions. Figure 2. Average Standard Deviation (%SD) with the variation between two of the six TNRD detectors represented as error bars, as a function of thermal neutron fluence (x105/cm2) in logarithmic representation. III. Thesis core 2. Publications: (A) A new online detector for estimation of perihpheral neutron equivalent dose in organ 58 III. Thesis core 2. Publications: (B.1) Using a Tandem Pelletron accelerator to produce a thermal neutron beam for detector testing purposes 59 (B.1) Using a Tandem Pelletron accelerator to produce a thermal neutron beam for detector testing purposes Irazola et al., Appl Radiat Isot 2016;107:330-334 Abstract Active thermal neutron detectors are used in a wide range of measuring devices in medicine, industry and research. For many applications, the long-term stability of these devices is crucial, so that very well controlled neutron fields are needed to perform calibrations and repeatability tests. A way to achieve such reference neutron fields, relying on a 3 MV Tandem Pelletron accelerator available at the CNA (Seville, Spain), is reported here. This paper shows thermal neutron field production and reproducibility characteristics over few days. 1. Introduction Achieving stable thermal neutron beams for calibrating and testing thermal neutron detectors is an important challenge in a number of fields where ionizing radiations are employed. Traditionally, metrology-grade thermal neutron fields are obtained by moderating radionuclide sources of 252Cf or 241Am–Be, with large polyethylene or graphite blocks. An example is the SIGMA facility of IRSN France (Muller et al.,2003; Lacoste et al.,2007). However, most of the existing facilities were decommissioned because their internal III. Thesis core 2. Publications: (B.1) Using a Tandem Pelletron accelerator to produce a thermal neutron beam for detector testing purposes 60 sources were too old to guarantee a safe operation. In addition, achieving large radionuclide sources for new facilities has became unfeasible for both economical and safety reasons. As a consequence, the scientific community is searching for alternative sources of thermal neutron fields. Exploiting the 7Li(p,n) reaction at near-threshold proton energies is a good option because established nuclear data are available and very low-energy neutrons can be achieved, thus requiring very reduced amount of additional thermalizing material. The 3 MV Tandem Pelletron accelerator1 3 (Praena et al.,2013) at CNA (Centro Nacional de Aceleradores, Sevilla, Spain) was used for this purpose. Some sheets of lead were added to reduce the photon field and a few cm of thick polyethylene moderator was adopted as moderator. The field was monitored, over three days of operation, using (a) a proton current integrator connected to the target backing, and (b) the active thermal neutron detector called TNRD (Thermal Neutron Rate Detector) (Bedogni et al.,2014), used in medical physics (Irazola et al.,2014; 2015b) to estimate neutron equivalent doses to peripheral organs for oncological patients treated with medical accelerators (Expósito et al.,2013), using the methodology established by Sánchez-Doblado et al.,2012, Gómez et al.,2010 and Romero-Expósito et al.,2015. 2. Material and method 2.1 TNRD neutron detector Figure 1 shows the TNRD detector, developed by INFN-LNF, Italy (Bedogni et al., 2014). This detector is based on a low-cost commercial solid-state device sensitized to thermal neutrons through a customized physical–chemical 1 http://www.pelletron.com/negion.htm III. Thesis core 2. Publications: (B.1) Using a Tandem Pelletron accelerator to produce a thermal neutron beam for detector testing purposes 61 Figure 1. One of the TNRD detectors (marked with a white box) and associated six electronics channels. treatment. Its active area is 1 cm2 and the overall dimensions are approximately 1.5 x 1 x 0.4 cm3 . It linearly responds in terms of thermal neutron fluence rate from 102 up to 106 cm-2 s-1. TNRD signal is amplified in a low-voltage electronics module and sent to a PC-controlled programmable ADC. Control software was developed in LabView© (2010 National Instruments). TNRD output is a DC voltage directly proportional to the thermal neutron fluence rate. Every TNRD is individually calibrated. The accuracy of the detector is within 75%, or better, for the fluence rate interval from 500 up to 106 cm-2 s-1 (Bedogni et al.,2014). Additional uncertainty terms should be added, in practical measurements, if the neutron field has unknown direction distribution and is superposed to an intense gamma component. The parasitic response of TNRD to photons has been additionally evaluated (Terrón et al.,2015; Irazola et al.,2015a,c). The III. Thesis core 2. Publications: (B.1) Using a Tandem Pelletron accelerator to produce a thermal neutron beam for detector testing purposes 62 reproducibility of TNRD, previously assessed using a constant thermal field from a moderated 241Am–Be source, is ±1.2% over a time interval of days. 2.2 Neutron spectrum determination The 7Li(p,n)7Be reaction has been studied in terms of total neutron yield, energy and angle distribution of the secondary neutrons as a function of the target thickness and projectile energy (Yu et al.,1998; Lederer et al.,2012). A FORTRAN code was written to generate the angle – and energy – distribution of neutrons based on analytical description of experimental data (Lee et al.,1999). MCNPX (v2.5) (Pelowitz et al.,2005) was used for their transport. ENDF/B-VII.0 and ENDF/ B-VI were used for particle-production and transport data and photoatomic data, respectively. The neutron spectra generated by 7Li(p,n)7Be at 1912 keV, which is used in the present experiment, was successfully modeled previously by Praena et al.,2013. The method of modelization, FORTRAN code for generation and MCNPX for transport, was also checked with neutron spectra emitted at different angles by the 7Li(p,n)7Be reaction near-threshold, Praena et al.,2014. MCNPX was also used to determine the optimal thickness of lead needed to reduce the parasitic photon field, due to the 477 keV photons from 7Li(p,n)7Be reaction. This value was fixed to 2.55 cm (17 lead sheets of 1.5 mm each), located 0.4 cm after the target. To thermalize the field, an optimized 2.2 cm thick polyethylene sheet was added immediately after the lead. Lateral size of both pieces was 20x20 cm2. TNRD detector was then placed in an aluminum support 3.5 cm after the polyethylene block (Praena et al.,2015). This is the conventional point of test. The complete setup is shown in Figure 2. Figure 3a shows simulated angle-integrated primary neutron spectrum at source position while Figure 3b displays the simulated neutron spectrum at detector position. It can be noticed that only neutrons of energy below 1 eV III. Thesis core 2. Publications: (B.1) Using a Tandem Pelletron accelerator to produce a thermal neutron beam for detector testing purposes 63 Figure 2. (a) Experimental setup aligned following the lithium target by using a fixed laser. (b) Detail of TNRD setup consisting on: a 2.55 cm lead layer and a 2.2 cm polyethylene (Ply) layer (both of 20x20 cm2) located between the lithium target and the detector (distance of approximately 8.9 cm). Inset shows TNRD location in the aluminum support. (a) TNRD detec tor Pb Ply laser TNRD setup (b) III. Thesis core 2. Publications: (B.1) Using a Tandem Pelletron accelerator to produce a thermal neutron beam for detector testing purposes 64 Figure 3. (a) Simulated neutron unitary spectrum: (a) at source position with 7Li(p,n) reaction at Ep=1912 keV and (b) at TNRD position using 2.2 cm polyethylene and 2.55 cm lead filters. Inset shows how the majority of neutrons below 1 keV have an energy lower than 1 eV. (a) (b) III. Thesis core 2. Publications: (B.1) Using a Tandem Pelletron accelerator to produce a thermal neutron beam for detector testing purposes 65 reach TNRD. Neutron spectra were obtained with a MCNP tally 4 which calculates de flux (n/cm2) averaged over the lithium target (a) TNRD detector (b) normalized to total number of neutrons generated in the simulation. 2.3 Proton accelerator and target The neutron beam was obtained from the CNA 3 MV Tandem Pelletron accelerator at 917 kV nominal voltage. Figure 4a shows the final part of the Basic Nuclear Physics (FNB) Tandem accelerator line, used in this experiment. It consists on a vacuum pipe housing a copper backing as cooling system. This backing holds a 50 μm thickness aluminum foil and the lithium target layer. The dimensions of this piece are 3x3x0.8 cm3 with a centered cylinder hole of 1 cm of diameter and 0.75 cm height, used to place the lithium layer (380 μm thickness). To prevent target melting, the copper support contains an internal cooling water circuit. Proton current on the lithium target was measured by connecting the copper backing to an Ortec Digital Current Integrator (Model 439). The nominal reproducibility of the electrometer is ±0.01%. Proton current could be varied up to about 2 μA, corresponding to a thermal neutron fluence of about 2500 cm-2 s-1 at the point of test. To prevent non-target contributions to the measured proton current, a double collimator system consisting of two rings, one in copper and the other in Teflon® (connected through two ceramic screws, as shown in Figure 4b), was used. Collimators and target holder have external diameter of 3 cm. Internal diameter is 1.1 cm for copper ring and 1 cm for Lithium target and Teflon® ring. Beam focusing (Figure 4c) was checked using a ViewPort2 4 (DN 40 CF) device coupled with a luminescent quartz screen. 2 https://www.pfeiffer-vacuum.com/productPdfs/420GSG040.en.pdf III. Thesis core 2. Publications: (B.1) Using a Tandem Pelletron accelerator to produce a thermal neutron beam for detector testing purposes 66 Figure 4. (a) Experimental setup at the 3 MV Tandem Pelletron accelerator at CNA (Seville), (b) copper-Teflon® disc used to estimate the current directly reaching the lithium target (blurred of the image is due to the fact that target has to be manipulated inside an Argon chamber to avoid lithium oxidation) and (c) monitor screen detail of the neutron beam collimated in the ViewPort. (a) (b) (c) III. Thesis core 2. Publications: (B.1) Using a Tandem Pelletron accelerator to produce a thermal neutron beam for detector testing purposes 67 3. Measurement results Measurements were performed during 3 different days. Every day, the accelerator setup (focalization, energy and proton current values) was fixed. In this phase, TNRD reading was observed as a function of the nominal proton energy, allowing to identify the reaction threshold (1880 keV) and to verify the energy calibration of the accelerator. Energy was then increased to the project value of 1912 keV. The proton current was tuned to achieve values of thermal neutron fluence in the order of 2x103 cm-2s-1. The three series of 15 min measurements performed in this condition are shown in Figure 5. For every measurement, the 15 min-time-integrated reading of the TNRD (termed TNRD) and of the proton current monitor (termed Q) were collected. The ratio between these two quantities, termed TNRDn, is the normalized TNRD reading. These quantities are reported in Table I for all measurements. The column s% reports the standard deviation of the measurements collected during each day. As expected, s% values for Q are slightly lower than for TNRD, meaning that the proportionality between proton current and thermal neutron fluence rate at the point is slightly perturbed by other beam-related sources of in fluence (positioning, focus, energy constancy). The impact of these sources of influence on the beam reproducibility may be estimated from the values of s% for TNRDn. These values, ±3.5%, ±2.2% and ±1.7%, have been corrected by subtracting in quadrature the TNRD reproducibility (±1.2%), obtaining ±3.3%, ±1.9% and ±1.2%. Every measurement day is characterized by a different average value of TNRDn (±1.24, ±1.31 and ±1.21), indicating that each time the accelerator is turned on and regulated, a slightly different point of work is achieved. Thus, the global inter-day uncertainty obtained for the thermal neutron beam is ±4.0%, taking into account TNRD reproducibility. The availability of a reliable thermal neutron monitor, in parallel to the proton current measuring device, will be a mandatory condition to achieve reproducible irradiation conditions on III. Thesis core 2. Publications: (B.1) Using a Tandem Pelletron accelerator to produce a thermal neutron beam for detector testing purposes 74 III. Thesis core 2. Publications: (B.2) Improving the neutron-to-photon discrimination capability of detectors used for neutron dosimetry in high energy photon beam radiotherapy 75 (B.2) Improving the neutron-to-photon discrimination capability of detectors used for neutron dosimetry in high energy photon beam radiotherapy Irazola et al., Appl Radiat and Isot, in press Abstract The increasing interest of the medical community to radioinduced second malignancies due to photoneutrons in patients undergoing high-energy radiotherapy, has stimulated in recent years the study of peripheral doses, including the development of some dedicated active detectors. Although these devices are designed to respond to neutrons only, their parasitic photon response is usually not identically zero and anisotropic. The impact of these facts on measurement accuracy can be important, especially in points close to the photon field-edge. A simple method to estimate the photon contribution to detector readings is to cover it with a thermal neutron absorber with reduced secondary photon emission, such as a borated rubber. This technique was applied to the TNRD (Thermal Neutron Rate Detector), recently validated for thermal neutron measurements in high-energy photon radiotherapy. The positive results, together with the accessibility of the method, encourage its application to other detectors and different clinical scenarios. III. Thesis core 2. Publications: (B.2) Improving the neutron-to-photon discrimination capability of detectors used for neutron dosimetry in high energy photon beam radiotherapy 76 1. Introduction New radiotherapy techniques, such as those with intensity modulation of the beam fluence, reduce the amount of healthy tissue exposed to high radiation doses. However, these techniques are usually associated to a greater demand in terms of Monitor Units (MU), which implies an increase of the out-of-field doses (also called peripheral doses; Xu et al.,2008). The latter, together with increment of the low-dose volumes and the larger survival after radiotherapy treatments, have made the incidence of late effects, such as second malignant neoplasms more relevant. Therefore, an important number of dosimetry studies have been conducted to determine peripheral doses more accurately. These doses have two main components: leakage/scattered photons and neutron contamination. Since dosimetric methods for photon doses are well-known (Sánchez-Nieto et al., 2015; Taddei et al.,2013), our group has focused on the neutron component, and established a methodology to estimate neutron peripheral doses (Sánchez-Doblado et al., 2012; Expósito et al., 2013, RomeroExpósito et al.,2015; Vázquez-Luque et al.,2013). This procedure was developed for a particular thermal neutron detector (Gómez et al., 2010) but applicable to any other (Guardiola et al., 2013; Bedogni et al., 2014). Specifically, the methodology by Sánchez-Doblado et al., 2012 was applied to a TNRD (Thermal Neutron Rate Detector) detector, designed and developed by Bedogni et al., 2014 and thus characterized for neutron peripheral dose measurements in radiotherapy environments. The TNRD showed satisfactory performances in terms of user friendliness and high sensitivity (Irazola et al.,2014). Nevertheless, recent experiments with the detector located ‘inphantom’ close to the border of the field, indicated the need for further investigations in relation to unexpected behaviors. A comprehensive analysis on the electronic, cable length and detector ageing, as possible causes for the rear events, was carried out. However, it turned out to be related with photon III. Thesis core 2. Publications: (B.2) Improving the neutron-to-photon discrimination capability of detectors used for neutron dosimetry in high energy photon beam radiotherapy 77 rejection issues and detector anisotropy under some critical conditions (Irazola et al., 205a;b;c; Praena et al.,2015; Irazola et al.,2016). These parasitic effects were indeed directly observed during exposures with 60Co and 6 MV Linac (Terrón et al.,2015), where no neutrons are present. However, a methodology for the neutron-to-photon discrimination in TNRD readings during high-energy photon radiotherapy has yet to be found. This paper proposes a simple method to estimate photon contribution to TNRD readings by covering the detector with a thermal neutron absorber (with reduced secondary photon emission), such as a borated rubber. For the present work a commercial material, called Flex-Boron® (http://www.deqtech.com/Shieldwerx/Data_Sheets/SWX-238.pdf), was used. This material (Gómez et al.,2010; D’Mellow et al.,2007) is expected to reduce incident thermal neutron field to less than 1%. Thus, the pure thermal neutron reading of a detector can be obtained by subtracting the reading of the rubber-covered detector to the uncovered one. 2. Material and method 2.1 TNRD detector TNRD detector, developed by Bedogni et al., 2014 in the framework of the NESCOFI@BTF project (2011-2013, Scientific Commission V, INFN-LNF, Italy), is based on a low-cost commercial solid-state device made sensitive to thermal neutrons through a customized physical–chemical treatment (mainly consisting on a 6Li deposition layer). Its active area is 1 cm2 and its overall dimensions are 1.5 cm x 1 cm x 0.4 cm (Figure 1). Its output is a DC voltage, which is proportional to the thermal neutron fluence rate (for this reason the device is called "rate detector"). This signal is amplified in a low-voltage electronics module especially developed by the project team. The amplified output is sent to a programmable ADC (NI USB-6218 BNC, 16 bit, up to 250 kilo samples per III. Thesis core 2. Publications: (B.2) Improving the neutron-to-photon discrimination capability of detectors used for neutron dosimetry in high energy photon beam radiotherapy 78 second) controlled by a PC through a LabView application. TNRD has a linear response to thermal neutron fluence rates exposures from 102 up to 106 cm-2·s-1. Every single TNRD (Figure 1) is individually calibrated by exposing it to a suitable reference thermal field (Bedogni et al.,2016). Detector-to-detector response variability is of the order of ±5% (1 SD). Figure 1. TNRD (Thermal Neutron Rate Detector). 2.2 The borated rubber Firstly, the photon absorption of borated rubber was characterized in relation to equivalent water material (a common surrogate of human tissue). Results showed that 0.32 cm (ρ=1.64 g/cm3) of rubber presented the same photon attenuation as 0.5 cm of polystyrene (ρ=1.05 g/cm3) usually employed. This equivalence is independent of the primary photon field energy, since the out-ofaxis photon spectrum is mainly composed by photons below 0.5 MeV (D’Mellow et al.,2007; Chofor et al.,2012). In addition, due to the high neutron capture cross section of 10B, the borated rubber acts as an efficient neutron absorber (Guardiola et al.,2013; http://www.johncaunt.com/shielding/neutronshielding/jc238/) ,with a nominal thermal neutron transmission factor of 3.8 x10-3. Two layers of Flex-boron® (5 cm x 20 cm) were used to cover above and below the set of five detectors used here. III. Thesis core 2. Publications: (B.2) Improving the neutron-to-photon discrimination capability of detectors used for neutron dosimetry in high energy photon beam radiotherapy 79 2.3 Irradiation tests Exposures were performed in a Siemens Primus Linac (6 and 15 MV) at Hospital Universitario Virgen Macarena, Seville (Spain) for a 40x10 cm2 field and 300 MU (1 MU delivers 1 cGy under reference conditions, e.g. 10x10 cm2, gantry 0º, source-axis distance 100 cm, depth of maximum for each energy in water). The following test were performed: (a) Free-in-air measurements During the irradiation of the patients, thermal neutron measurements are performed with bare detectors located in front of the couch, at about 3 m distance (Expósito et al.,2013; Figure 2). Neutron peripheral dose estimations are done from detector readings following the methodology established in Sánchez-Doblado et al., 2012. Under this conditions, it is expected a minimal photon contribution to the detector reading and a low directionality dependence for the detector. This assumption was tested under an adverse clinical scenario representative of a breast treatment. It consisted of four beams of high (15 MV) and low (6 MV) energy, applied alternatively using 122° and 300° incidence angles for each energy. 8 MU and 168 MU for the high and low energy beams were used, respectively. (b) In-phantom measurements at different distances Out-of-field photon doses were determined with a PTW Farmer® ion chamber 30013 (operated at -250 V) and a PTW Unidos electrometer. Measurements were carried out at a set of positions from 0.1 to 2 m from the border of the field, with 0º gantry incidence (perpendicular to the treatment couch). Ion chamber and/or five TNRD detectors were inserted in a set of plastic layers (30x30 cm2) simulating ‘in-phantom’ attenuation; patient scatter was mimicked with additional plastic material (Figure 3). III. Thesis core 2. Publications: (B.2) Improving the neutron-to-photon discrimination capability of detectors used for neutron dosimetry in high energy photon beam radiotherapy 80 Figure 2. TNRD location for patient measurements (free-in-air). Two different settings inside the (8 cm thick) phantom were used: one where the ion chamber and the TNRD detectors were placed together at mid depth of the plastic phantom (insert i in Figure 3) and the other one where the TNRD detectors were placed (insert ii in Figure 3) between borated rubber layers (Flexboron®) with the equivalent replaced plastic thickness (see section 2.2). “Plastic” and “borated” measurements were carried out at low energy (6 MV), where neutron presence is negligible and at high energy (15 MV). (c) In-phantom measurements at different beam incidence angles In order to evaluate TNRD anisotropic response, two specific field-edge distances (0.15 and 0.35 m) were measured for 14 different beam angle incidences, covering 360°. III. Thesis core 2. Publications: (B.2) Improving the neutron-to-photon discrimination capability of detectors used for neutron dosimetry in high energy photon beam radiotherapy 81 Figure 3. (a) Set up for the photon sensitivity study for measurements outside the treatment field (0.1 to 2 m from the field edge) for five of the devices. Additional plastic material has been used to simulate patient scatter. TNRD detectors were placed as shown in insert (i) with Flexboron material above and below and (ii) with an ionization chamber location for photon dose estimations when plastic is used. Figure (b) represents a schema of detector location in the slab phantom. 3. Results and discussion 3.1 Free-in-air measurements Figure 4a shows the time-dependent TNRD reading for the selected clinical scenario. Detector "baseline" is represented by the red dashed line. The α and β peaks correspond to the neutron signal from the 15 MV beam. Inserts show a zoom of TNRD readings when: (4b) no beam is present and (4c) only the 6 MV beam is on (negligible neutron presence). The behavior of TNRD readings showed in insert (4c), indicates that photons induce baseline oscillations but, as expected, their average is null in practice. It should be noted that, due to the anisotropic response of the detector, the point where a photon-induced secondary electron is generated would affect the III. Thesis core 2. Publications: (B.2) Improving the neutron-to-photon discrimination capability of detectors used for neutron dosimetry in high energy photon beam radiotherapy 82 Figure 4. (a) TNRD signal during the free-in-air measurement for a real breast treatment case, combining 6&15 MV. Baseline is represented by the red dashed line. Insert (b) and (c) show a zoom of TNRD signal for no beam and 6 MV beam, respectively. charge collection event, which may result in a positive or negative pulse. The internal structure of the detector was designed so that the convolution of these signals yields a zero-average voltage level in photon fields. By contrast, in neutron fields, the average level is a positive value proportional to the fluence rate. This TNRD behavior makes it valid during patient irradiation (free-in-air measurements) where photon contribution is slight compared with neutron one. However, problems in photon rejection can become more important when photon background increases as during “inphantom” measurements (see later on). III. Thesis core 2. Publications: (B.2) Improving the neutron-to-photon discrimination capability of detectors used for neutron dosimetry in high energy photon beam radiotherapy 83 3.2 In-phantom measurements at different distances Figure 5 reports TNRD readings for "plastic" and "borated" measurements with 6 MV. Photon doses at each measurement point were known as detectors were exposed together with the ion chamber (see 2.3a). As expected, in absence of neutrons, both readings coincide within uncertainties. That is, borated rubber does not introduce perturbation to photon attenuation. Similarly, Figure 6 shows "plastic" and "borated" measurements in 15 MV. The line represents neutron signal (obtained as the following subtraction: "plastic" - "borated"). As expected, the parasitic photon signal gains relevance at short distances (≤ 0.3 m from the field-edge). Percentage of photon to total signal it ranges from 3.7% at 2 m up to 26% at 0.15 m. 3.3 In-phantom measurements at different beam incidence angles Figure 7a shows TNRD "borated" readings for fourteen different beam incidences (Figure 7b) at two border of the field to detector distances. As expected from previous experiences (Irazola et al.,2015c), photon component compensates for complementary angles. Thus, its influence is almost negligible if “balanced” incidence angles (uniformly covering 360º) are used. This is a usual approximation used in conventional radiotherapy treatments that was also used in Irazola et al.,2014. There is an important angular photon dependence of TNRD devices, being clearly different for beam incidences comprised in the range (280º-75º) or (250º-105º). This problem would be solved with the use of the proposed methodology. However, when bare detectors are used for these measurements in “balanced” treatments, due to the almost complementary TNRD response, the parasitic averaged photon contribution found was -4.4% at 0.15 m and +1.5% at 0.35 m, being lower for further points. III. Thesis core 2. Publications: (B.3) Neutron measurements in radiotherapy: a method to correct neutron sensitive devices for parasitic photon response 90 treatment energy. Measurements with ionization chamber and Monte Carlo simulations were used to evaluate the validity of this hypothesis. DEPS method was compared to already published correction methods, such as the use of neutron absorber materials. In addition to its simplicity, an advantage of DEPs procedure is that it can be applied to any radiotherapy machine. 1. Introduction New radiotherapy (RT) techniques, such as those based on beam fluence intensity modulation, are known to provide higher healing rates and larger survival after treatments. However, this increase in life expectancy has brought to light some late effects such as second malignant neoplasms. Although these techniques provide a greater degree of conformity, they are also known to increase peripheral doses (PD) due to their higher demand of Monitor Units (MU) and to dose escalation strategies, usually associated to these modern techniques (Xu et al.,2002). PD are mainly composed of two different contributions: leakage/scattered photons and neutrons. While photon peripheral doses have been widely investigated (Van der Giessen et al.,2001; Sánchez-Nieto et al., 2015; Jagetic et al.,2015), neutron contamination has been largely undervalued. In previous works, this group established a method to infer peripheral neutron doses, estimated from in-phantom thermal neutron measurements. (Gómez et al.,2012; Sánchez-Doblado et al.,2012; Expósito et al.,2013; Romero-Expósito et al.,2015). These neutron doses have been used to assess secondary cancer risks. On this basis, different treatment strategies can be compared and precious contribution can be given to the process of choosing the best treatment for a given clinical case. III. Thesis core 2. Publications: (B.3) Neutron measurements in radiotherapy: a method to correct neutron sensitive devices for parasitic photon response 91 A recently developed thermal neutron detector named TNRD (Thermal Neutron Rate Detector, Bedogni et al.,2014), was successfully used for these in-phantom thermal neutron measurements (Irazola et al.,2014). The detector is based on solid-state devices which thermal neutron response is enhanced using 6Li radiators. Although the detector was designed to minimize the response to photons, this effect cannot be neglected when the measurement point is located near the field-edge, i.e. under intense photon background (Terrón et al.,2015; Irazola et al.,2015c). Overestimating thermal neutron fluence in peripheral organs could lead to systematic errors in calculating secondary cancer risks, with potential impact on the clinical decisions. Thus, a thorough study was performed to better understand these effects on the TNRD reading (Irazola et al., 2015a; Praena et al., 2015; Irazola et al., 2016; Irazola et al., 2015b; Terrón et al., 2015; Irazola et al., 2015c). The method proposed in this work, called “Double Energy Photon Subtraction” (DEPS), derives from the hypothesis that photon doses to peripheral points are energy independent in the megavoltage range (Mazonakis et al.,2008). Under this assumption, the pure photon contribution to the TNRD reading could be estimated from measurements at a treatment energy where no (or very little) neutrons are produced, such as 6 MV. The same treatment delivered at higher energy will thus produce an over-reading in the detector, due to neutron contribution. Clearly, the machine calibration must be the same for both energies. 2. Material and method 2.1 TNRD detector TNRD detector (Bedogni et al., 2014) is based on a low-cost commercial solidstate device made sensitive to thermal neutrons through a customized physical– chemical treatment. Its active area is 1 cm2 and its overall dimensions are 1.5 cm III. Thesis core 2. Publications: (B.3) Neutron measurements in radiotherapy: a method to correct neutron sensitive devices for parasitic photon response 92 x 1 cm x 0.4 cm. TNRD output is a DC voltage, which is proportional to the thermal neutron fluence rate (for this reason the device is called "rate detector"). This signal is amplified in a specifically developed low-voltage electronic board. The amplified output is sent to a programmable ADC (NI USB-6218 BNC, 16 bit, sampling rate up to 2.5x105 samples per second) controlled by a PC through a LabView application developed by the Politecnico di Milano. TNRD linearly responds to thermal neutron fluence rates from 102 up to 106 cm-2·s-1. Every detector is calibrated in terms of thermal neutron fluence in a reference thermal field (Bedogni et al., 2016). Detector-to-detector response variability is in the order of ±5% (1 SD). An additional uncertainty term of about 10%, applicable in clinical conditions, was estimated by Irazola et al., 2014. 2.2 DEPS Method The DEPS method is based on the hypothesis that photon doses to peripheral points are energy-independent in the megavoltage range (Mazonakis et al.,2008). Under this assumption, the pure photon contribution to the TNRD reading could be estimated from measurements at a treatment energy where no (or very little) neutrons are produced, such as 6 MV (hereafter called "lowenergy" beam). The same treatment delivered at higher energy (15 MV in this work) will thus produce an over-reading in the detector, due to neutron contribution. Clearly, the machine calibration must be the same for both energies. This energy-independence of PPD was tested by Monte Carlo simulations and measurements, as described below. 2.2.1 Monte Carlo simulations Monte Carlo simulations have been used to compute mean energy of the photon spectra for both nominal energies, from the isocenter to 55 cm. EGSnrc Monte Carlo user code BEAMnrc (Kawrakow et al., 2011; Rogers et al., 2011) was used to simulate 6 and 15 MV photon beams from a Siemens Primus linac. III. Thesis core 2. Publications: (B.3) Neutron measurements in radiotherapy: a method to correct neutron sensitive devices for parasitic photon response 93 The following BEAMnrc/EGSnrc transport parameters were employed: NIST for bremsstrahlung cross sections; EXACT as boundary crossing algorithm and PRESTA-II as electron-step algorithm. For bremsstrahlung angular sampling, the leading term of Koch-Motz distributions was chosen; electron and photon cutoff energies were 0.512 MeV (0.001 MeV kinetic energy) and 0.001 MeV, respectively. Electron range rejection, with an energy cutoff of 2.0 MeV, was implemented. Bremsstrahlung splitting was activated to obtain the first Phase Space Data (PSD) file for the 6 MV case. The PSD files corresponding to each aperture were obtained below the MLC for a 10x10 cm2 field in a plane located 10 cm depth (SSD=100 cm) in a water block (60x60x30 cm3). The number of primary histories launched from the source in the PSD was 5 and 50 million for the 6 and 15 MV cases respectively, chosen to satisfy a spatial density of 105 particles/cm2 and trying to avoid latent uncertainty in order to obtain an adequate level of statistical accuracy. For out-of-field energy distribution comparison, the fact that simulations were performed with the information provided by the manufacturer (incomplete in terms of external shielding) could result in a minor overestimation that is not expected to distort our results. As the out-of-field statistics are lacking, several simulations with different initial seeds were performed. The mean photon energy was regarded as a relevant quality index for the photon spectra in peripheral points. Simulated mean energies at different distances from the field-edge can be used to identify the proper calibration factor for ionization chambers (IC). 2.2.2 Farmer ionization chamber measurements An IC based dosimetry system (PTW 30013 IC operating at -250 V and a PTW UNIDOS® electrometer) was used. Measurements for lowand high-enegy beams were taken by varying the field-edge distance from 0.1 to 2 m. The IC was allocated in a plastic insert, sandwiched between 8 cm of polystyrene and III. Thesis core 2. Publications: (B.3) Neutron measurements in radiotherapy: a method to correct neutron sensitive devices for parasitic photon response 94 with additional blocks of plastic mimicking patient scatter. Irradiations were performed in a Siemens Primus linac for a 40x10 cm2 field (in order to fully cover the 6 available TNRD devices, accounting for scattering), delivering 300 MU (dose rate of 300 MU·min-1) with gantry angle at 0º. 2.2.3 Irradiations with TNRD TNRD measurements at lowand high-energy, were performed under the same setup and irradiation conditions as those previously described for the IC (section 2.2.2). The photon contribution to TNRD readings at a peripheral point was measured using the 6 MV beam. Then, neutron contribution to TNRD readings, TNRD-n, was calculated by subtracting the low-energy reading from the high-energy one. To validate the DEPS method, TNRD-n values were directly compared with those obtained with a different method adopting a Flex-boron® filter to remove the thermal neutron component (Irazola et al.,2016). 3. Results and discussion 3.1 Monte Carlo simulations: results Figure1 depicts mean photon energy computed by Monte Carlo simulations. At the isocenter the mean energy is 1.4 and 3.2 MeV for the 6 and 15 MV fields, respectively. At peripheral points (≥10 cm to the field edge), the mean photon energy is 280 keV with ±10% maximum variation when the beam is changed from 6 MV to 15 MV. The corresponding variability in the mass energy absorption coefficient (μen/ρ) for Silicon is smaller than ±1.1%, implying no effect on the TNRD photon response (Terrón et al., 2015). The energy response of the IC is also flat in this energy range (Shani, 2000; Aird et al.,1972; Andreo et al.,2006). III. Thesis core 2. Publications: (B.3) Neutron measurements in radiotherapy: a method to correct neutron sensitive devices for parasitic photon response 95 Figure 1. Photon mean energy of the spectra, in water, computed by Monte Carlo for distances ranging from 0 to 55 cm far from the isocenter in a Siemens Primus linac in 6 and 15 MV. 3.2 Results from Farmer IC measurements Figure 2 shows the readings of the IC at 6 and 15 MV as the field-edge distance varies from 0.1 to 2 m. Readings have been normalized to the 15 MV case at 0.1 m. The corresponding absorbed dose for 300 MU ranges from 0.07 cGy (at 2 m) to 4.00 cGy (at 0.1 m). The plot experimentally demonstrates that PPD does not depend on beam energy. Distance to the isocenter (cm) 010 20 30 40 50 60 Mean Energy (MeV) 0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 6 MV 15 MV III. Thesis core 2. Publications: (B.3) Neutron measurements in radiotherapy: a method to correct neutron sensitive devices for parasitic photon response 96 Figure 2. Normalized Farmer ionization readings obtained for the studied field edge-detector distances (0.1 to 2 m) in 6 and 15 MV. 3.3 TNRD results Figure 3 shows TNRD signals at 6 and 15 MV for the studied field-edge distances. For the high-energy case, both photons and neutrons contribute to the reading of the detector (solid dots) whereas for the low-energy beam TNRD only sees photons (hollow triangles). TNRD readings in 6 MV can be subtracted from those of the 15 MV case (containing photon and neutron signals), in order to obtain the neutron contribution (TNRD-n). Figure 4 compares the TNRD-n values with those obtained with a different method, adopting a Flex-boron® filter to remove the thermal neutron Distance to the field edge (m) 0,0 0,5 1,0 1,5 2,0 Normalized values 0,0 0,2 0,4 0,6 0,8 1,0 1,2 15 MV 6 MV III. Thesis core 2. Publications: (B.3) Neutron measurements in radiotherapy: a method to correct neutron sensitive devices for parasitic photon response 97 Figure 3. Global TNRD measurement in high (photon+neutron represented as solid dots) and low (photon, hollow triangles) energies for the studied distances. All measurements refer to 300 MU. TNRD axis readings correspond to thermal neutron fluences up to around 18 x106 n·cm2, depending on detector calibration factor. Distance to the field edge (m) TNRD (V·s) 0 20 40 60 80 100 120 140 160 15 MV - 15 MV absorber 15 MV6 MV 0.5 1 1.5 2 0 Figure 4. Neutron contribution to TNRD reading at 15 MV obtained by Flex-Boron (black) or DEPS method (white). All measurements refer to 300 MU. TNRD axis readings correspond to thermal neutron fluences up to around 16 x106 n·cm2, depending on detector calibration factor. TNRD (V·s) 0 20 40 60 80 100 120 140 160 180 15 MV (photon+neutron) 6 MV (photon) Distance to the field edge (m) 00.2 0.4 0.6 0.8 2 11.2 1.4 1.6 1.8 III. Thesis core 2. Publications: (B.3) Neutron measurements in radiotherapy: a method to correct neutron sensitive devices for parasitic photon response 98 component (Irazola et al.,2016). In the latter case, a pair of exposures at 15 MV are performed for every field-edge distance: one with the TNRD covered with Flex-boron®, the other uncovered. Again, the neutron contribution is obatined by subtraction. Differences are smaller than uncertainties of the estimations, proving that the two methods are equivalent. 4. Conclusions This work proposes a simple method to correctly estimate the thermal neutron reading of active thermal neutron detectors used ‘in-phantom’ for peripheral neutron dose experiments in high-energy RT. Despite these detectors are designed to reject photons, their photon response cannot be neglected when the measurement point is near to the field-edge. The method is called “Double Energy Photon Subtraction” (DEPS) and relies on the hypothesis that photon doses to peripheral points are energy independent in the megavoltage range. Under this assumption, the pure photon contribution to the detector reading is estimated from measurements at a treatment energy where no (or very little) neutrons are produced, such as 6 MV. Measurements at low (6 MV) and high (15 MV) energies are performed under the same setup. The neutron contribution to detector reading is obtained by subtracting the low energy reading from the high energy one. Monte Carlo simulations, as well as ion chamber measurements were used to validate the basic assumptions of DEPS method. This was applied to the recently developed thermal neutron detector TNRD. Experiments were performed by varying the field-edge distance from 0.1 to 2 m. The results were succesfully compared with those obtained by another method, based on FlexBoron filters. III. Thesis core 2. Publications: (B.3) Neutron measurements in radiotherapy: a method to correct neutron sensitive devices for parasitic photon response 99 The DEPS approach represents a simple and universal correction method, usable at any high energy facility equipped with low energy (i.e. 6 MV). It offers an operative procedure without the need of any extra specific material, setup modifications for ‘in-phantom’ accommodation of the detectors and avoiding the necessary dismantling and re-assembling of this setup for each pair series of measurements. 5. References Aird EG and Farmer FT. The design of a Thimble Chamber for the Farmer Dosemeter. Phys Med Biol 1972;17(2):169-174. Andreo P, Burns DT, Hohlfeld K, Saiful Huq M, Kanai T, Laitano F, Smth V and Vynckier S. IAEA TRS-398. Absorbed Dose Determination in External Beam Radiotherapy: An International Code of Practice for Dosimetry based on Standards of Absorbed Dose to Water. IAEA 2006. Bedogni R, Bortot D, Pola A, Introini MV, Gentile A, Esposito A, Gómez-Ros JM, Palomba M, and Grossi A. A new active thermal neutron detector. Radiat Prot Dosim 2014;161(1–4):241–244. Bedogni R, Sacco D, Gómez-Ros JM, Lorenzoli M, Gentile A, Buonomo B, Pola A, Introini MV, Bortot D and Domingo C. ETHERNES: A new design of radionuclide source-based thermal neutron facility with large homogeneity area. Appl Radiat and Iso 2016;107:171–176. Chofor N, Harder D and Poppe B. Non-reference condition correction factor kNR of typical radiation detectors applied for the dosimetry of high-energy photon fields in radiotherapy. Z Med Phys 2012;22:181-196. Expósito MR, Sánchez-Nieto B, Terrón JA, Domingo C, Gómez F and SánchezDoblado F. Neutron contamination in radiotherapy: Estimation of second cancers based on measurement in 1377 patients. Radiother and Oncol 2013;107:234-241. Gómez F, Iglesias A and Sánchez-Doblado F. A new active method for the measurement of slow-neutron fluence in modern radiotherapy treatment rooms. Phys Med Bio. 2010;55:1025-1039. Irazola L, Lorenzoli M, Bedogni R, Pola A, Terrón JA, Sánchez-Nieto B, Expósito MR, Lagares JI, Sansaloni F and Sánchez-Doblado F. A new online detector for estimation of peripheral neutron equivalent dose in organ. Med Phys 2014;41:112105. III. Thesiscore 2. Publications: C. Neutron model upgrade for peripheral neutron dose assessment evaluated in 510 radiotherapy patients 106 establish a reasonable assessment of low dose risk estimates in radiotherapy (Diallo et al.,2009). The methodology proposed by our group (Gómez et al.,2010; Sánchez-Doblado et al.,2012; Expósito et al.,2013) provides a systematic estimation of neutron organ-equivalent doses at 12 organs for patients undergoing high energy external photon beam radiotherapy. However, despite the simplicity of this method, it requires the characterization of the facility in terms of neutron production by a specific SRAM-based detector, or the use of general values tabulated for each linac model (Sánchez-Doblado et al.,2012). In consequence, a more universal methodology would be desirable for this procedure. This model enables the consideration of patient sex and treatment location, namely head and neck (H&N) or abdomen. Once methodology was clinically validated (Expósito et al.,2013), we thought that a more general procedure being also more patient-specific would be highly desirable for clinical routine. Second Cancer Probability (SCP) estimation represents a step further in peripheral dose studies and some uncertainties remain regarding these magnitudes based on data from the Japanese atomic bomb survivors and the RBE of neutrons (Newhauser et al., 2016). However, recent studies have established the pass from neutron dose to risk by the direct use of tabulated values (Expósito et al.,2013), we have focused this study in the enhancement of neutron dose assessment, which would allow a direct estimation of SCP by only applying these factors, choosing the more convenient protocol. Former to the improvement of the existing models, we wanted to ensure that these two model locations were general enough to cover real treatment ones. For that, the goodness of these models was previously evaluated in real treatments, giving good concordance with TNRD estimations (Irazola et al.,2016). III. Thesiscore 2. Publications: C. Neutron model upgrade for peripheral neutron dose assessment evaluated in 510 radiotherapy patients 107 The goal of this work is the improvement of the previously existing methodology for PND estimation to organ, in order to make it more global and customized. We aimed to generate further patient-specific (regarding anatomy) neutron organ dose models that can be used in any facility, by terms of a simple characterization of linac neutron production, which can be easily performed with any thermal neutron detector. These new models will be used to compare calculated and measured peripheral neutron organ-equivalent doses for 510 patients, evaluated against results obtained with the old methodology. 2. Material and method A thermal neutrons-sensitive detector called TNRD (Thermal Neutron Rate Detector) was used. This neutron detector was initially designed by INFN-LNF (Italy; Bedogni et al.,2014) for nuclear purposes. The device is based on a lowcost commercial solid-state device sensitized to thermal neutrons through a customized physical-chemical treatment, with overall dimensions of approximately 1.5x1x0.4 cm3 (active area of 1 cm2). Further details on its composition can be found in (Bedogni et al.,2014). Those detectors respond linearly to any thermal neutron fluence rate in the range of 10 up to 106 cm-2·s-1 neutrons and show detector-to-detector response variability of the order of ±5% (1SD). However, for clinical environments an additional source of uncertainty, coming from beam angular entry and energy dependences had to be considered. Therefore, a global uncertainty of 11% was estimated by Irazola et al.,2014 for these devices in radiotherapy environments. Encompassing this value with procedure own incertitude, a global uncertainty of around 15% was establish for model generation methodology. One advantage of the TNRD detectors is their online-features, which greatly ease the processes of model improvement for which repeated measurements have to be usually made. Additionally, the reduced size of these active III. Thesiscore 2. Publications: C. Neutron model upgrade for peripheral neutron dose assessment evaluated in 510 radiotherapy patients 108 detectors allows not only ‘external’ (as used during the development of the neutron equivalent dose estimation model in Sánchez-Doblado et al.,2012) but also and ‘in-phantom’ measurements. TNRD thermal neutron fluence estimations are obtained in compliance with ISO 8529-1. 2.1 Facility neutron fluence characterization The original methodology (Sánchez-Doblado et al.,2012) relied on the correlation between neutron production (expressed as single events upset (SEU) of a SRAM-based neutron detector) and the neutron equivalent dose to peripheral organs. The latter required the neutron detector to be present during patient irradiation. Romero-Expósito et al.,2015 proposed a generalization of that methodology by characterizing the facility, in terms of photoneutron production, so that the presence of the detector was not required posteriorly during patient irradiations. For this, the previous determination, under certain reference conditions, of a characterization parameter (in units of events per MU, termed c) was required. The former parameter, escalated by the MU used during specific irradiations, correlated with neutron equivalent dose to organs. This neutron characterization should be done just once (e.g., during linac commissioning). A further step in the generalization of the neutron dose estimation methodology involves the characterization of the facility in terms of thermal neutron fluence. Therefore, the implementation of the neutron dosimetry methodology in any facility could be carried out by characterizing it by means of any thermal neutron detector. This involves the estimation of the characterization parameter (c*) in units of neutrons per cm2 and MU using eq.1 with the neutron detector outside the III. Thesiscore 2. Publications: C. Neutron model upgrade for peripheral neutron dose assessment evaluated in 510 radiotherapy patients 109 beam (in front of the gantry and close to the bunker wall) under reference conditions (i.e., gantry angle at 0º, field size of 10×10 cm2) as proposed in (Romero-Expósito et al.,2015). 𝑐∗=𝛷𝑡ℎ·𝐹𝐵 𝑀𝑈 (eq.1) where: - 𝛷𝑡ℎrepresents the accumulated thermal neutron fluence in the reference location and irradiation conditions; - MU is the total number of high energy MU delivered (usually 1000 MU) - FB is a factor accounting for the ratio from the bunker volume, with a floor area of A (m2), to a reference one with a floor area of 64.6 m2. This factor was modeled as 𝐹𝐵=1 (0.46+34.13/𝐴) ⁄ calculated from that described in in (Sánchez-Doblado et al.,2012) In the present work, the TNRD detectors will be used for the implementation of this upgraded methodology together with the incorporation of patient’s size. 2.2 In-phantom measurements Three anthropomorphic NORMA phantom models (Figure 1) were available to cover several patient ages (child teen and adult). Phantoms were manufactured in polyethylene whereas low-density wood was used for simulating the lung tissue. 16 customized detector holes were distributed along each phantom, at different depths, representing relevant cancer-at-risk organ locations for the estimation of equivalent dose in organs. Table I lists the point(s) used for organ dose investigation, extended from that of Sánchez-Doblado et al.,2012. Firstly, a validation of the use of the model by Sánchez-Doblado et al.,2012, with TNRD was carried out for the adult phantom. Afterwards, the III. Thesiscore 2. Publications: C. Neutron model upgrade for peripheral neutron dose assessment evaluated in 510 radiotherapy patients 110 generalization of the existing model to account for different patient sizes was carried out. The same irradiation conditions as those of the original model (SánchezDoblado et al.,2012) were used: (i) Abdomen treatment: eight beam incidences (at 0º, 45º, 90º, 135º, 180º, 225º, 270º, 315º), 10x10 cm2 field sizes, 15 MV, isocentre at point A (Figure 1) (ii) Head and neck treatment: the same beam incidences, field sizes and energy than in (i), isocentre at point H (Figure 1) Irradiations were set to 1000 monitor units (MU) with a dose rate of 300 MU·min-1 and were carried out with a Siemens Primus linac. 2.2.1 Model generalization for its use with thermal neutron detectors In order to use of TNRDs detectors in the methodology established by SánchezDoblado et al.,2012 a comparison between thermal neutron fluences measured with passive detectors (TLDs and PACD, Poly Allyl Diglycol Carbonate) during model development and those of TNRDs was carried out for the adult phantom. Previous to this comparison, passive detector fluence values were corrected to account for the differences in the cut-off energies used during the calibration procedures for both types of detectors (Irazola et al.,2014). Model in Sánchez-Doblado et al.,2012, encompassed average values for a wide range of available combinations of linacs and energies; thus, model estimations were associated to an uncertainty of around 30%. It will be assumed that agreement, within uncertainties, of both sets of fluence measurements will imply that the generalized model (in terms of thermal neutron fluence) is also applicable to any linac. III. Thesiscore 2. Publications: C. Neutron model upgrade for peripheral neutron dose assessment evaluated in 510 radiotherapy patients 111 The agreement between these results and those of the previous models, together with the previously generalized characterization procedure to be performed in terms of thermal neutron fluence (eq.1), allows the use of this methodology with any thermal neutron detector, instead of being expressed in terms of events (Sánchez-Doblado et al.,2012). Table I. NORMA´s optimized detector locations specific to 14 radiosensitive internal organs. Modified from Sánchez-Doblado et al.,2012, adding two extra organs*. k Organ NORMA Points 1 Thyroid 4 2 Oesophagus 4,9,16 3 Lung 7,8 4 Breast 5,6,15 5 Stomach 9,11,16 5 Liver 9,10,11,16 6 Colon 11,12 7 Urinary Bladder 10 8 Ovary 11,12 9 Skin 15 10 Bone surface 1,3,9,12,13,14,15 11 Marrow 9,12,15 12 Remainder All except 7,8,15 13 Prostate* 11,12 14 Uterus* 11,12 III. Thesiscore 2. Publications: C. Neutron model upgrade for peripheral neutron dose assessment evaluated in 510 radiotherapy patients 112 Figure 1. NORMA Phantom models to cover adult, teen and child sizes and points locations specific to 14 cancer-at-risk organs (Table I). Figure modified from Sánchez-Doblado et al.,2012; González-Soto et al.,2012. III. Thesiscore 2. Publications: C. Neutron model upgrade for peripheral neutron dose assessment evaluated in 510 radiotherapy patients 113 2.2.2 Model generalization to account for patient size (a) Thermal neutron fluence measurements for the three phantom sizes The already available Thermal neutron fluences (Sansaloni et al.,2011), measured with TLDs, were to be confirmed with the new TNRD devices for the three phantoms and irradiation sets above described (for abdomen and H&N locations) in the 16 points. Both uncertainties were estimated at 15%, as measurements were carried out using the same linac. (b) Organ-equivalent neutron dose model The previously mentioned methodology (Sánchez-Doblado et al.,2012) is based on the convolution at the 16 points of the normalized Monte Carlo neutron fluence energy spectra with the kerma factor for ICRU tissue (k) (Siebert et al.,1995) and energy-dependent radiation weighting factor from ICRP-103 (wR). The same convolution approach to estimate the neutron-equivalent dose in an organ k, estimated from average of measured points i of a patient undergoing a treatment j (Sánchez-Doblado et al.,2012; Romero-Expósito et al.,2016), but using directly the thermal neutron fluence (Φth), is now proposed as follows: 𝐻𝑖,𝑗 =𝛷𝑖,𝑗 𝑡ℎ ·𝐹𝑖,𝑗 (eq.2) where: - 𝛷𝑖,𝑗 𝑡ℎ is the accumulated thermal neutron fluence (n·cm-2), corrected by bunker size parameter (FB, firstly introduced in eq.1), at any point (i) for any treatment (j) - Fi,j are the neutron-equivalent dose transformation factors, for each NORMA point i and for type of treatment j, considering total neutron III. Thesiscore 2. Publications: C. Neutron model upgrade for peripheral neutron dose assessment evaluated in 510 radiotherapy patients 114 spectra 𝜑𝑖,𝑗(𝐸), radiation weighting factors 𝑤𝑅(𝐸) and kerma factor (k(E)). Therefore, eq.2 can be used to estimate the neutron-equivalent dose in an organ k (from the averaged values of measured i-points) of a patient undergoing a jtreatment from the thermal neutron fluence at that point (provided, in this case, by the TNRDs detector). In order to introduce the facility characterization (section 2.1) in the expression, the {𝛷𝑖,𝑗 𝑡ℎ} matrix data were correlated to thermal neutron fluences measured with the TNRD detector located at reference position in the bunker, R (SánchezDoblado et al.,2012). If Mi,j represent those correlation coefficients, eq.3 can be then rewritten as follows: 𝐻𝑖,𝑗(µ𝑆𝑣)=𝑅·𝑀𝑖,𝑗 (eq.3) where: - R (x106 n·cm-2) is the accumulated thermal neutron fluence at the reference location inside the bunker during a treatment. In practice, R value can be estimated using either: o the characterization approach (Rcal = c*·MU) which makes use of the specific MU used for the irradiation, o or directly measuring thermal neutron fluence at the reference location during the irradiation (Rmeas) and correcting it by bunker surface factor (𝐹𝐵). - Mi,j (µSv·10-6 n-1·cm2) represent the corresponding Hi,j values from eq.3 per thermal fluence of the reference detector (establishing a correlation among dose equivalent at every i-point and reference thermal neutron measurement in the treatment room). It allows the direct estimation of neutron dose equivalent at III. Thesiscore 2. Publications: C. Neutron model upgrade for peripheral neutron dose assessment evaluated in 510 radiotherapy patients 115 the studied i-points for the desired j-treatment location by the only knowledge of thermal neutron contamination in the reference location for the studied treatment. In order to evaluate organ-equivalent neutron doses to the 12 k-organs detailed in Table I, we used the same point to organ aggrupation than those established in Sánchez-Doblado et al.,2012 to go from the Mi,j dose-to-point factors to the Mk,j organ-to-dose factors that govern the model (correlating organ-equivalent dose to reference thermal neutron fluence in the room, R). (c) Organ-equivalent neutron dose model considering patient dimensions Finally, in order to consider different patient anatomy (height and weight), the Body Surface Area (BSA) concept, introduced by Du Bois et al.,1916, was considered. Accordingly, the height and weight of a patient were combined in a single BSA parameter (termed HW from now onwards) as follows: 𝐻𝑊 (𝑚2)=0.007184·ℎ0.725·𝑤0.425 (eq.4) where h and w stand for patient height (cm) and weight (kg) respectively. Then, the HW parameter was used as a surrogate for mapping for each patient distances form the organ to the source as well as the depth of the i-points, corresponding to each k-organ. The later mainly affects to the neutron spectra at the point. Thus, the neutron equivalent dose to the k-organ during the j-treatment can be calculated from the following equation: [𝐻]𝑇 𝑘,𝑗(𝜇𝑆𝑣)=𝑅·[𝐸(𝐻𝑊)]𝑘,𝑗 (eq.5) where: III. Thesis core 2. Publications: (C) Neutron model upgrade for peripheral neutron dose assessment evaluated in 510 radiotherapy patients 122 (c) Organ-equivalent neutron dose model considering patient dimensions The HW parameter for each phantom size was calculated so that the functions E(HW)k,j describing the Mk,j values as a function of the HW parameter were estimated for the abdomen and H&N as follows: [𝐸]𝑘,𝑎𝑏𝑑 =10−3· [ (−214.1𝐻𝑊+563.9)𝑡ℎ𝑦𝑟𝑜𝑖𝑑 (−309.2𝐻𝑊+900.7)𝑜𝑒𝑠𝑜𝑝ℎ𝑎𝑔𝑢𝑠 (−834.9𝐻𝑊2+2420𝐻𝑊−808)𝑙𝑢𝑛𝑔 (1496𝐻𝑊2−5153.8𝐻𝑊+5314)𝑏𝑟𝑒𝑎𝑠𝑡 (−253.4𝐻𝑊+753.3)𝑠𝑡𝑜𝑚𝑎𝑐ℎ (−228.3𝐻𝑊+755.7)𝑙𝑖𝑣𝑒𝑟 (−146.3𝐻𝑊+359.4)𝑐𝑜𝑙𝑜𝑛 (−152.9𝐻𝑊+762.8)𝑏𝑙𝑎𝑑𝑑𝑒𝑟 (−146.3𝐻𝑊+360)𝑜𝑣𝑎𝑟𝑦 (3133.5𝐻𝑊2−9453.6𝐻𝑊+9019.1)𝑠𝑘𝑖𝑛 (581.4𝐻𝑊2−1916.1𝐻𝑊+2042.2)𝑏𝑜𝑛𝑒 (1235𝐻𝑊2−3911.1𝐻𝑊+4000.1)𝑚𝑎𝑟𝑟𝑜𝑤 (−330𝐻𝑊+814.2)𝑟𝑒𝑚𝑎𝑖𝑛𝑑𝑒𝑟 (−146.3𝐻𝑊+360)𝑝𝑟𝑜𝑠𝑡𝑎𝑡𝑒 (−146.3𝐻𝑊+360)𝑢𝑡𝑒𝑟𝑢𝑠 ] 𝑎𝑛𝑑 [𝐸]𝑘,𝐻&𝑁 =10−3· [ (−128.3𝐻𝑊+618.4)𝑡ℎ𝑦𝑟𝑜𝑖𝑑 (−200.1𝐻𝑊+641.6)𝑜𝑒𝑠𝑜𝑝ℎ𝑎𝑔𝑢𝑠 (−356.6𝐻𝑊+1149.9)𝑙𝑢𝑛𝑔 (−844.8𝐻𝑊+2046.9)𝑏𝑟𝑒𝑎𝑠𝑡 (−168.4𝐻𝑊+460.1)𝑠𝑡𝑜𝑚𝑎𝑐ℎ (−172.9𝐻𝑊+470.4)𝑙𝑖𝑣𝑒𝑟 (−95.7𝐻𝑊+202.6)𝑐𝑜𝑙𝑜𝑛 (−186.5𝐻𝑊+501.4)𝑏𝑙𝑎𝑑𝑑𝑒𝑟 (−95.73𝐻𝑊+202.6)𝑜𝑣𝑎𝑟𝑦 (1726.7𝐻𝑊2−5281.3𝐻𝑊+4649.6)𝑠𝑘𝑖𝑛 (319.06𝐻𝑊2 − 1091.1𝐻𝑊 + 1192.8)𝑏𝑜𝑛𝑒 (700.71𝐻𝑊2 − 2272.9𝐻𝑊 + 2186.3)𝑚𝑎𝑟𝑟𝑜𝑤 (−280.3𝐻𝑊+715.1)𝑟𝑒𝑚𝑎𝑖𝑛𝑑𝑒𝑟 (−95.73𝐻𝑊+202.6)𝑝𝑟𝑜𝑠𝑡𝑎𝑡𝑒 (−95.73𝐻𝑊+202.6)𝑢𝑡𝑒𝑟𝑢𝑠 ] Those expressions are to be used in eq.5 for organ-equivalent dose estimation. III. Thesis core 2. Publications: (C) Neutron model upgrade for peripheral neutron dose assessment evaluated in 510 radiotherapy patients 123 Figure 4 depicts a selection of the qualitative behavior found for the fitting functions. Figure 4. Selection of some representative fitting curves for Abdomen and H&N cases considering the patient size parameter (HW) for: (a) Thyroid and (b) Prostate (similar to the fitting behavior obtained for oesophagus, stomach, liver, colon, bladder, ovary, remainder and uterus); (c) Skin (similar to breast, bone and marrow) and special case of (d) Lung. As expected, for the majority of the organs, the equivalent dose decreases when patient size increases, either due to a higher height (implying further distance from the source) or to a greater weight (entailing higher neutron thermalization and decrease in the neutron fluence). However skin (or organs containing this point for calculation as breast, bone and marrow) showed a different pattern. That behavior can be explained trough the two confounding factors. In small patients (i.e., HW < 1.5 m2), the III. Thesis core 2. Publications: (C) Neutron model upgrade for peripheral neutron dose assessment evaluated in 510 radiotherapy patients 124 higher dose is mainly due to the greater fast-neutrons component which decreases with the inverse square law of the distance to the source. In patients with HW > 1.5 m2 the fast neutron component should be lower but seems to be compensated by the higher surface exposed to neutron radiation. Equivalent dose to lungs during abdominal irradiation also shows a particular dependence with patient size. During this irradiation part of the lungs lays inside the treatment field (isocenter in point A, Figure 1), which has been shown to be representative of treatments in the abdominal area (Irazola et al.,2016a). The measured TNRD thermal neutron fluences showed Figures 3a, 3b a,d 3c are in agreement with MC simulations (Figure 5). The final behavior of the fitting curve can be understood as a result of the balance of the thermal and fast neutron component. Thus, starting from different behavior of the neutron fluence components: thermal (increasing with) and fast (decreasing with phantom size). This option is especially useful for retrospective studies for which not all the patient information (height or weight) is available, using as a first approximation patient age to choose the most adequate phantom size. This would improve peripheral neutron dose estimation, especially in the case of child patients. Figure 5. Monte Carlo simulation of neutron spectra for abdomen treatment location for the three phantom sizes (González-Soto et al.,2012), lethargy representation. III. Thesis core 2. Publications: (C) Neutron model upgrade for peripheral neutron dose assessment evaluated in 510 radiotherapy patients 125 3.2 Patient estimations Figure 6 depicts average organ-equivalent neutron doses estimated with the original and the generalized (eq.5) models for 240 patients in three organs. Those organs (thyroid, liver and marrow) were just selected for illustrative purposes. Differences among these values represent improvements in thermal neutron estimations at some points thanks to TNRD capabilities and the added modification of including patient size (HW). As a general behavior previous values seemed to overestimate the dose (in agreement with results from Figure 2). In order to evaluate the goodness of the Mk,j values for retrospective studies (or other cases where there is no information about patient size), measurements and theoretical estimations for organ-equivalent neutron doses for the 510 studied patients were compared. As all the patients are adult-aged, Mk,j adult values (abdomen and H&N models) were chosen for this purpose. Results are plotted in Figure 7. As expected, as all patients included in the study were adult, no high deviations were found between methodologies. Nevertheless, greater deviations should be expected for young patients due to the higher differences in height and weight from the adult NORMA phantom. For the chosen example (Figure 8), the old model underestimated in average the organequivalent doses by 34.1% (assuming that the new model provides the correct patient-size estimation). III. Thesis core 2. Publications: (C) Neutron model upgrade for peripheral neutron dose assessment evaluated in 510 radiotherapy patients 126 Figure 6. Comparison between organ-equivalent neutron doses estimated with the old (white) and the new (black) models for 240 patients in three organs: (a) Thyroid, (b) liver and (c) marrow for the four studied pathologies. III. Thesis core 2. Publications: (C) Neutron model upgrade for peripheral neutron dose assessment evaluated in 510 radiotherapy patients 127 Figure 7. Comparison of measurement and theoretical estimations (Mk,j factors for adult) for organ-equivalent doses for the 510 studied patients. III. Thesis core 2. Publications: (C) Neutron model upgrade for peripheral neutron dose assessment evaluated in 510 radiotherapy patients 128 H (mSv) 0 2 4 6 8 10 12 14 New Model Old Model Thy. Oes. Lung Bre. Stom. Liv. Col.Blad.Ova. Skin Bone Marr. Rem. Prost. Uter. Figure 8. Comparison of peripheral doses for an abdomen location child treatment case calculated with the old model and the new one (considering patient size). Concerning SCP, as we are working in low-dose range (e.g. <2.5 Gy), linear models could be used for this purpose (Schneider et al.,2005). Thus, eq.5 can be employed for SCP estimates by using tabulated coefficients from ICRP-103 or BEIR-VII (sexand age-at-exposurespecific) protocols: 𝑆𝐶𝑅=𝐻𝑘,𝑗 ·𝜆𝑘 (eq.6) where 𝐻𝑘,𝑗 are peripheral neutron doses for organ -k in the specific model -j and 𝜆𝑘 are the organ risk factors from ICRP-103 or BEIR-VII protocols. III. Thesis core 2. Publications: (C) Neutron model upgrade for peripheral neutron dose assessment evaluated in 510 radiotherapy patients 129 4. Conclusions The use of the new online thermal neutron miniaturized active TNRD detectors, for peripheral neutron dose assessment, has allowed the improvement of existing models for organ-equivalent neutron dose estimation in any facility. Several aspects have been enhanced: - Characterization of the facility has been established in terms of thermal neutron fluence instead of the events of the previously used SRAMbased detector, generalizing the methodology to any thermal neutron detector. - The online behavior of TNRD detector, has simplified the procedures followed for model enhancements. - Models have been modified to be more patient-specific, introducing weight and height to better evaluate organ-equivalent neutron doses, which results especially relevant for children cases. A universal methodology for peripheral neutron organ-equivalent doses estimation has been established. It only requires the characterization of the facility in terms of thermal neutron fluence, the total number of high energy MU delivered and patient size (height and weight). Finally these models have been validated by the comparison of estimations from the existing models to a cohort of 510 patients. This would be useful for future second cancer risk evaluation and its implementation in TPS. 5. References Armstron GT, Kawashima T, Leisenring W, Stratton K, Stovall M, Hudson MM , et a. Aging and risk of severe, disabling, life-htretening, and fatal eventsin the childhood cancer rurvivor study. J Clin Oncol 2014;32:1218-27. Bedogni R, Bortot D, Pola A, Introini MV, Gentile A, Esposito A, Gómez-Ros JM, Palomba M, Grossi A. A new active thermal neutron detector. Radiation Protection Dosimetry.Radiat Prot Dosim 2014;161( 1–4)241–244. III. Thesis core 2. Publications: (C) Neutron model upgrade for peripheral neutron dose assessment evaluated in 510 radiotherapy patients 130 Bedogni R, Sacco D, Gómez-Ros JM, Lorenzoli M, Gentile A, Buonomo B, Pola A, Introini MV, Bortot D and Domingo C. ETHERNES: A new design of radionuclide source-based thermal neutron facility with large homogeneity area. Appl Radiat and Iso 107 (2016) 171–176. BEIR Phase 2,NRC National Research Council. Health risks from exposure to low levels of ionizing radiation., 2006. Berdnarz B, Hancox C and Xu XG. Calculated organ doses from selected prostate treatment plans using Monte Carlo simulations and an anatomically realistic computational phantom. Phys Med Biol 2009;54:5271-86. Chargari C, Goodman KA, Diallo I, Guy JB, Rancoule C, Cosset JM, Deutsch E and Magne N. Risk of second cancers in the era of modern radiation therapy: does the risk/benefit analysis overcome theoretical models? Cancer Metastasis Rev 2016;113. Diallo I, Haddy N, Adidj E Samand A, Quiniou E, Chavaudra J, Alziar I, Perret N, Guérin S, Lefkopoulos D and de Vathaire F. Frequency distribution of second cancer locations in relation to the irradiated volume among 115 patients treated for childhood cancer. Int J Rad Oncol Biol Phys 2009;74 (3):876-883. Dubois D, Dubois EF. A formula to estimate the approximate surface area if height and weight be known. Arch Intern Med 1916; 17:863-871. Expósito MR, Sánchez-Nieto B, Terrón JA, Domingo C, Gómez F and Sánchez-Doblado F. Neutron contamination in radiotherapy: Estimation of second cancers based on measurement in 1377 patients. Radiother and Oncol 2013;107:234-241. Followill D, Nüsslin F and Orton CG. IMRT should not be administered at photon energies greater than 10 MV. Med Phys 2007;34(6):1877-1879. Gómez F, Sánchez-Doblado F, Iglesias A and Domingo C. Active on-line detector for inroom radiotherapy neutron measurements. Radiat Meas 2010;45:1532-1535. Gónzalez-Soto X, Expósito M., Sánchez-Nieto B, Amgarou K, Lagares JI, Gómez F, Domingo C and Sánchez-Doblado F. Neutron spectra inside an adult and children anthropomorphic phantoms in high energy radiotherapy. In: Long M, editors WC 2012, IFMBE Proccedings 39, Berlin:Springer. 2012:1145-1148. Hall EJ and Wuu CS. Radiation-induced second cancers: The impact of 3D-CRT and IMRT. Int J Radiation Oncology Biol Phys 2003;56(1):83-88. Harrison RM. Introduction to dosimetry and risk estimation of second cancer induction following radiotherapy. Radiat Meas 2013;57:1-8. Howell RM, Hertel NE, Wang Z, Hutchinson J, Fullerton GD. Calculation of effective dose from measurements of secondary neutron spectra and scattered photon dose from dynamic MLC IMRT for 6 MV, 15 MV and 18 MV beam energies. Med Phys 2006;33:360-8. ICRP International Commission on Radiological Protection. The 2007 Recommendations of the International Commission on Radiological Protection. ICRP Publication 103, 2007. Irazola L, Lorenzoli M, Bedogni R, Pola A, Terron JA, Sanchez-Nieto B, Exposito MR, Lagares JI, Sansaloni F and Sanchez-Doblado F. A new online detector for estimation of peripheral neutron equivalent dose in organ. Med Phys 2014;41:112105. Irazola L, Terrón JA, Bedogni R, Lorenzoli M, Pola A, Sánchez-Nieto B and SánchezDoblado F. Signal photon component of a new thermal neutron detector TNRD in radiotherapy environments. Radiother and Oncol 2015;115(1):S870. III. Thesis core 2. Publications: (C) Neutron model upgrade for peripheral neutron dose assessment evaluated in 510 radiotherapy patients 131 Irazola L, Terrón JA, Sánchez-Nieto B, Romero-Expósito M and Sánchez-Doblado F. Peripheral neutron dose model verification for real IMRT cases. Poster at the 1st European Congress of Medical Physics (ECMP) 2016, Athens; Greece (2016a) Irazola L, Terron JA, Bedogni R, Pola A, Lorenzoli M, Sanchez-Nieto B, Gomez F and Sanchez-Doblado F. Limitations and solutions of TNRD neutron detector in high energy photon beam radiotherapy. Sent to Appl Radiat Isot (2016b) ISO Standard, TC 85 Nuclear energy, nuclear technologies, and radiological protection SC 2. Reference neutron radiations – Part 1: Characteristics and methods of production. ISO 8529-1;2001. Jagetic L and Newhauser WD. A simple and fast physics-based analytical method to calculate therapeutic and stray doses from external beam, megavoltage x-ray therapy. Phys Med Biol 2015;60:4753-4475. Kry SF, Slehpour M, Followill DS, Stovall M, Kuban DA, White RA and Rosen II. Outof-field photon and neutron dose equivalents from step-and-shoot intensity-modulated radiation therapy. J Radiation Oncology Biol Phys 22005;62:1204-16. Nath R, Boyer A, LaRiviere PD, McCall RC and Price KW. Neutron measurements around high energy X-ray radiotherapy machines. A Report of Task Group 27, Radiation Therapy Committee, American Asociation of Physicists in Medicine. AAPM Report No.19. Medical Physics Publishing, NY, 1986. Newhauser WD and Durante M. Assessing the risk of second malignancies after modern radiotherapy. Nat Rev Cancer 2011;11(6):439-448. Newhauser WD, Berrington de Gonzalez A, Schulte R and Lee C. A Review of Radiothearpy-Induced Late Effects Research after Advanced Technology Treatments. Front Oncol 2016;6:13. NCRP-79, National Council on Radiation Protection and Measurements. Neutron contamination from medical electron accelerators. Reperot No. 79. National Council on Radiation Protection and Measurements, Bethesda, MD, 1984. Pena J, Franco L, Gómez F, A Iglesias, J Pardo, and M Pombar. Monte Carlo study of siemens PRIMUS photoneutron production. Physics in Medicine and Biology, 50:5921–5933, 2005. Romero-Expósito M, Sánchez-Nieto B, Terrón JA, Lopez MC, Ferreira BC, Grishchuk D, Sandíns C, Moral-Sánchez S, Bragado-Álvarez L, Melchor M, Domingo C, Gómez F and Sánchez-Doblado F. Comminsioning the neutron production of a linac: development of a clinical planning tool for second cancer risk estimation. Med Phys 2015;42:276-281. Romero-Expósito M, Domingo C, Sánchez-Doblado F, Ortega-Gelabert O and Gallego S. Experimental evaluation of neutron dose in radiotherapy patients: which dose? Med Phys 2016;43:360. Romero-Expósito M, Sánchez-Nieto B, Irazola L and Sánchez-Doblado F. In regard to “Neutron contamination in radiotherapy: Estimation of second cancers based on measurements in 1377 patients.” Letter to the Editor. Sent to Radiother and Oncol Ruben JD, Davis S, Evans C, Jones P, Gallgiardi F, Haynes M and Hunter A. The effect of intensity-modulated radiotherapy on radiation-induced second malignancies. J Radiation Oncology Biol Phys 2008;70(5):1530-1536. Sánchez-Doblado F, Domingo C, Gómez F, Sánchez-Nieto B, Muñiz JL, García-Fusté MJ, Expósito MR, Barquero Terrón JA, et al. Estimation of neutron equivalent dose in