Browsing by Author "Espinoza Bornscheuer, Ignacio Guillermo"

Now showing 1 - 11 of 11
  • Thumbnail Image
    Item
    A critical analysis of Shuryak’s Predictive Radiocarcinogenesis Model
    (2024) Heumann Schröder, Nicolás Matthias; Sánchez Nieto, Beatriz; Espinoza Bornscheuer, Ignacio Guillermo; Pontificia Universidad Católica de Chile. Instituto de Física
    Objective: To critically analyze, simplify and implement a predictive radiocarcinogenesis model to estimate the risk of secondary cancer after RT which can effectively compare different radiotherapy treatment plans with the aim of having an additional element of information during the decision-making process for the best RT plan.Methodology: A Python software was developed that was able to implement the model proposed by Shuryak et al. (2009). Simplifications and minor corrections were made which allowed for more compactness and more efficient run times. The model was then reparametrized with newer data from the Surveillance, Epidemiology, and End Results Program (SEER) database and several epidemiological studies using Bayesian Inference. Uncertainty propagation studies were then conducted to understand their propagation better. Finally, the model with its new parameters was applied to a selection of prostate plans to determine if it could construct a risk hierarchy.Results: The model was successfully reparametrized with newer data. Although some parameters show significant deviation from Shuryak’s original parameters, they are mostly on the same order of magnitude, and the differences arise likely due to differences in fitted data and the fitting process itself. Shuryak’s model successfully built a risk hierarchy between prostate plans, although it deviated from the more simplistic linear non-threshold BEIR VII model. It was also possible to simplify some complex mathematical equations, both in general and for particular cases, allowing for easier implementation and more efficient run times.Conclusions: Shuryak’s model was successfully reparametrized and implemented, showing potential to become clinically applicable. However, more comparisons between the model’s result and epidemiological data must be made to evaluate its accuracy better, and more concise and complete second primary cancer studies must be used before the model is reliable enough for clinical decision-making.
  • Thumbnail Image
    Item
    A dynamic nonlinear coupled differential equations model of tumor growth and response to radiotherapy
    (2018) Ortiz Guzmán, Valentina; Gago Arias, Araceli; Espinoza Bornscheuer, Ignacio Guillermo; Karger, Christian; Pontificia Universidad Católica de Chile. Facultad de Física
  • Thumbnail Image
    Item
  • Thumbnail Image
    Item
    A simple analytical model for a fast 3D assessment of peripheral photon dose during coplanar isocentric photon radiotherapy
    (2022) Sánchez Nieto, Beatriz; López Martínez, Ignacio N.; Rodríguez Mongua, José Luis; Espinoza Bornscheuer, Ignacio Guillermo
    Considering that cancer survival rates have been growing and that nearly two-thirds of those survivors were exposed to clinical radiation during its treatment, the study of long-term radiation effects, especially secondary cancer induction, has become increasingly important. To correctly assess this risk, knowing the dose to out-of-field organs is essential. As it has been reported, commercial treatment planning systems do not accurately calculate the dose far away from the border of the field; analytical dose estimation models may help this purpose. In this work, the development and validation of a new three-dimensional (3D) analytical model to assess the photon peripheral dose during radiotherapy is presented. It needs only two treatment-specific input parameter values, plus information about the linac-specific leakage, when available. It is easy to use and generates 3D whole-body dose distributions and, particularly, the dose to out-of-field organs (as dose–volume histograms) outside the 5% isodose for any isocentric treatment using coplanar beams [including intensity modulated radiotherapy and volumetric modulated arc therapy (VMAT)]. The model was configured with the corresponding Monte Carlo simulation of the peripheral absorbed dose for a 6 MV abdomen treatment on the International Comission on Radiological Protection (ICRP) 110 computational phantom. It was then validated with experimental measurements using thermoluminescent dosimeters in the male ATOM anthropomorphic phantom irradiated with a VMAT treatment for prostate cancer. Additionally, its performance was challenged by applying it to a lung radiotherapy treatment very different from the one used for training. The model agreed well with measurements and simulated dose values. A graphical user interface was developed as a first step to making this work more approachable to a daily clinical application.
  • Thumbnail Image
    Item
    A voxel-based multiscale model to simulate the radiation response of hypoxic tumors
    (2015) Espinoza Bornscheuer, Ignacio Guillermo
  • Thumbnail Image
    Item
    Can non-propagating hydrodynamic solitons be forced to move?
    (2011) Gordillo, L.; Espinoza Bornscheuer, Ignacio Guillermo
  • Thumbnail Image
    Item
    Diagnóstico de cáncer pulmonar a través de un análisis cuantitativo de imágenes de tomografía computarizada.
    (2020) Paredes Ahumada, Juan Antonio; Espinoza Bornscheuer, Ignacio Guillermo; Pontificia Universidad Católica de Chile. Facultad de Física
    El trabajo realizado consistió en el desarrollo de un código computacional capaz de encontrar diferencias estadísticamente significativas entre imágenes de tomografía computacional en poblaciones de pacientes con nódulos pulmonares benignos y malignos, obtenidos del repositorio abierto TCIA [2]. Los features utilizados para diferenciar las poblaciones consistieron en features morfológicos tridimensionales, features de primer orden y basados en matriz de co-ocurrencia del nivel de gris (GLCM). Estos features se obtuvieron con el software MITK [3] luego de segmentar el nódulo pulmonar mediante Region growing en la misma plataforma. Con los features que mostraron diferencias entre ambas poblaciones se construyeron modelos univariados de Machine Learning con los algoritmos Regresión logística (LR), Vecinos cercanos (KNN), Arbol de decisión ´ (DT), Máquinas de soporte vectorial (SVM) y Naive Bayes (NB) para predecir la clasificación de los nódulos pulmonares, eliminando aquellos features correlacionados que tenían un bajo rendimiento. Se evaluó la métrica F1 para analizar el desempeño de los modelos, utilizando validación cruzada con k = 5 repetida 20 veces. Se realizó también un modelo multivariado buscando la combinación de features que tiene la menor razón de falsos negativos, encontrando como resultado los features Promedio, Disimilitud, Compacticidad y Correlación con el algoritmo NB, obteniéndose un FNR = 0.307 ± 0.036 y AUC = 0.678 ± 0.025. Este conjunto de features están en concordancia con relaciones previamente encontradas [4,5] y con características que el profesional médico observa comúnmente en imágenes de tomografía computarizada.
  • Thumbnail Image
    Item
    Dosimetría pre-tratamiento con dispositivos electrónicos de imagen portal (EPID)
    (2022) López Espinoza, Analiz; Espinoza Bornscheuer, Ignacio Guillermo; Caprile Etchart, Paola F.; Pontificia Universidad Católica de Chile. Facultad de Física
    Una de las principales ventajas de los dispositivos electrónicos de imagen portal (EPID) es que tienen una muy buena resolución espacial (0.25 mm/píxel). Si bien su uso fue primeramente para verificación del posicionamiento del paciente antes de su sesión de radioterapia, hoy en día es un elemento importante en la verificación de las dosis de tratamiento a ser administradas al paciente. El objetivo de este trabajo fue examinar las características dosimétricas de estos dispositivos, e implementar un modelo matemático que permita obtener dosis absorbida en un plano de un fantoma homogéneo de agua sólida, a partir de imágenes de transmisión de planes pre- tratamiento. Luego de una amplia búsqueda bibliográfica, se usó como referencia el trabajo de Wendling et al (2006), al cual se le agregaron las siguientes modificaciones: una corrección por sensibilidad de píxeles de la imagen PSM, el cálculo del factor de dispersión del paciente que llega al EPID y la ecuación del kernel de dispersión en el plano del fantoma homogéneo. Los resultados de dosis absorbida obtenidos fueron satisfactorios, con un error porcentual de 1.53% como máximo en el eje central, y en la comparación de perfiles de campos cuadrados (no modulados) pasaron el criterio Gamma del 3% de dosis y 3 mm en DTA en el 100% de puntos evaluados. Sin embargo, en la verificación de planes de tratamiento los resultados no fueron satisfactorios, pero nos permitió establecer las calibraciones físicas que deben realizarse al panel para utilizarlo con fines dosimétricos, no solo la dosimetría absoluta y relativa del haz de tratamiento, además de la importancia en la configuración correcta del modo de adquisición de imágenes para emplearlas con fines dosimétricos. Ya que los resultados para campos cuadrados tuvieron resultados aceptables, podemos sugerir que este modelo sirve para realizar controles de calidad diarios del linac donde, se emplean campos estáticos no modulados, con cierta UM y tasa de dosis fija.
  • Thumbnail Image
    Item
    Impact of different biologically-adapted radiotherapy strategies on tumor control evaluated with a tumor response model
    (2018) Gago Arias, Araceli; Sánchez Nieto, Beatriz; Espinoza Bornscheuer, Ignacio Guillermo; Christian P. Karger,; Pardo-Montero, Juan
  • No Thumbnail Available
    Item
    Simulation of hypoxia PET-tracer uptake in tumours : Dependence of clinical uptake-values on transport parameters and arterial input function
    (2020) Paredes Cisneros, I.; Karger, C. P.; Caprile Etchart, Paola F.; Nolte, D.; Espinoza Bornscheuer, Ignacio Guillermo; Gago Arias, Araceli
  • Thumbnail Image
    Item
    Study of out-of-field dose in photon radiotherapy: A commercial treatment planning system versus measurements and Monte Carlo simulations
    (2020) Sánchez Nieto, Beatriz; Medina Ascanio, Karem Nathalie; Rodríguez Mongua, José Luis; Doerner, Edgardo; Espinoza Bornscheuer, Ignacio Guillermo
    Purpose: An accurate assessment of out-of-field dose is necessary to estimate the risk of second cancer after radiotherapy and the damage to the organs at risk surrounding the planning target volume. Although treatment planning systems (TPSs) calculate dose distributions outside the treatment field, little is known about the accuracy of these calculations. The aim of this work is to thoroughly compare the out-of-field dose distributions given by two algorithms implemented in the Monaco TPS, with measurements and full Monte Carlo simulations. Methods: Out-of-field dose distributions predicted by the collapsed cone convolution (CCC) and Monte Carlo (MCMonaco) algorithms, built into the commercially available Monaco version 5.11 TPS, are compared with measurements carried out on an Elekta Axesse linear accelerator. For the measurements, ion chambers, thermoluminescent dosimeters, and EBT3 film are used. The BEAMnrc code, built on the EGSnrc system, is used to create a model of the Elekta Axesse with the Agility collimation system, and the space phase file generated is scored by DOSXYZnrc to generate the dose distributions (MCEGSnrc). Three different irradiation scenarios are considered: (a) a 10 x 10 cm(2)field, (b) an IMRT prostate plan, and (c) a three-field lung plan. Monaco's calculations, experimental measurements, and Monte Carlo simulations are carried out in water and/or in an ICRP110 phantom. Results: For the 10 x 10 cm(2)field case, CCC underestimated the dose, compared to ion chamber measurements, by 13% (differences relative to the algorithm) on average between the 5% and the approximate to 2% isodoses. MC(Monaco)underestimated the dose only from approximately the 2% isodose for this case. Qualitatively similar results were observed for the studied IMRT case when compared to film dosimetry. For the three-field lung plan, dose underestimations of up to approximate to 90% for MC(Monaco)and approximate to 60% for CCC, relative to MC(EGSnrc)simulations, were observed in mean dose to organs located beyond the 2% isodose. Conclusions: This work shows that Monaco underestimates out-of-field doses in almost all the cases considered. Thus, it does not describe dose distribution beyond the border of the field accurately. This is in agreement with previously published works reporting similar results for other TPSs. Analytical models for out-of-field dose assessment, MC simulations or experimental measurements may be an adequate alternative for this purpose.