Publication: Estimación automática de la edad ósea mediante procesado y segmentación de radiografías
Loading...
Identifiers
Publication date
2011-10
Defense date
2011-10-06
Authors
Advisors
Tutors
Journal Title
Journal ISSN
Volume Title
Publisher
Impact
Export
Abstract
A través del presente proyecto se pretende desarrollar un nuevo método estimatorio de la edad en un ser humano partiendo de una radiografía, utilizando diversas técnicas de segmentado de huesos junto el cálculo de distancias y su relación con la edad mediante un regresor lineal. Con independencia de los progresos ya alcanzados en este sentido, el concepto de un método automático supondrá un avance médico, por ejemplo en la medicina forense o en pediatría, sólo para citar algunos ejemplos. Los actuales métodos sobre segmentación de imágenes disponibles actualmente, y en particular el más común de ellos, se explican a lo largo de este estudio. La idea principal de este nuevo método se basa en el hecho de que el hueso está en continuo crecimiento hasta el final de la adolescencia, tendiendo a unirse en torno a los demás y provocando una reducción de la distancia entre ellos. Principalmente la aplicación de estas técnicas de segmentación para separar los huesos del resto de la imagen, será suficiente para etiquetar cada región, permitiendo obtener la distancia entre ellos, así como una estimación fidedigna de la edad. Sin embargo, este nuevo método presentado requerirá un proceso largo y difícil debido, ante todo, a la falta de información al respecto. Han sido exploradas diferentes metodologías: el primer enfoque se basa en los bordes que delimitan cada uno de los huesos, mientras que el segundo propone una hipótesis fundamentada en una mezcla de modelos que segmente las diferentes regiones, como por ejemplo una distribución de Gauss. Precisamente, la estimación de la edad está directamente relacionada en la segmentación y la medida de las diferentes distancias entre todos los huesos. Para lograr el resultado óptimo, el método debe ser completamente automático y funcional en todos sus términos. Además del estudio de viabilidad de estas técnicas, las principales conclusiones son las ventajas y desventajas de cada procedimiento, y también la comparación en términos eficaces y de precisión entre los diferentes métodos. Con el fin de obtener dicha información con facilidad, se desarrollará un algoritmo completo basado principalmente en la distancia entre los píxeles de la imagen, y será desarrollada una aplicación intuitiva compuesta por una sencilla interfaz gráfica, la cuál ofrece una forma interactiva y directa de trabajar con la imagen. ______________________________________________________________________________________________________________________________
The objective of this project is to develop a new method to estimate the age of a human being from a simple radiograph using some bone estimation techniques and a minimum distance criteria. Despite the great progress that has been previously achieved in this regard, the whole concept and theory behind the automatic method could suppose a medical breakthrough, as for example in Forensics or Pediatrics, just to name a few. The current methods available with the same purpose and the most common in particular, are explained through along this study. The main idea of this new method is based on is the fact that as the bone grows by it tends to join around other ones what results in a reduction of the distance between them. In a few words, the application of some segmentation techniques, basically to tell the bones apart from the rest of the image, will be enough to label each one, let to calculate the distance between them and allows obtaining age estimation therefore. However, this new method will be a hard and long-term process due to the lack of information about it. Different methodologies have been explored: the first approach is based on the edges which delimit each bone, while the second one proposes an hypothesis based on a mixing-model to segmentate the different regions, as for example a Gaussian distribution. Precisely, the age estimation is directly related on this segmentation and the measure of the different distances between all the bones. To achieve the optimal result, the method must be fully automatic and functional at all their terms. In addition to the feasibility study of these techniques, the main conclusions include the advantages and weaknesses of each procedure, and some comparatives about the efficiency between the different methods as well. In order to obtain such information easily, it becomes useful to develop a complete algorithm based mainly on the distance between certain pixels of the image, and also developing an intuitive application with a simple graphic user interface which offers an interactive and direct way to work with the rendered image.
The objective of this project is to develop a new method to estimate the age of a human being from a simple radiograph using some bone estimation techniques and a minimum distance criteria. Despite the great progress that has been previously achieved in this regard, the whole concept and theory behind the automatic method could suppose a medical breakthrough, as for example in Forensics or Pediatrics, just to name a few. The current methods available with the same purpose and the most common in particular, are explained through along this study. The main idea of this new method is based on is the fact that as the bone grows by it tends to join around other ones what results in a reduction of the distance between them. In a few words, the application of some segmentation techniques, basically to tell the bones apart from the rest of the image, will be enough to label each one, let to calculate the distance between them and allows obtaining age estimation therefore. However, this new method will be a hard and long-term process due to the lack of information about it. Different methodologies have been explored: the first approach is based on the edges which delimit each bone, while the second one proposes an hypothesis based on a mixing-model to segmentate the different regions, as for example a Gaussian distribution. Precisely, the age estimation is directly related on this segmentation and the measure of the different distances between all the bones. To achieve the optimal result, the method must be fully automatic and functional at all their terms. In addition to the feasibility study of these techniques, the main conclusions include the advantages and weaknesses of each procedure, and some comparatives about the efficiency between the different methods as well. In order to obtain such information easily, it becomes useful to develop a complete algorithm based mainly on the distance between certain pixels of the image, and also developing an intuitive application with a simple graphic user interface which offers an interactive and direct way to work with the rendered image.
Description
Keywords
Proceso de señales, Proceso de imágenes, Técnicas de segmentación de imágenes, Radiografías