Desarrollo de una aplicacion software para la caracterizacion geometrica de lesiones cutaneas/lunares potencialmente peligrosas
Archivos
NO SE HA AUTORIZADO la consulta de los documentos asociados
Fecha
2017-02-24
Autores
Título de la revista
ISSN de la revista
Título del volumen
Editor
Jaén: Universidad de Jaén
Resumen
[ES] El TFG consta del desarrollo de una aplicacion en Matlab para la deteccion semiautomatica de lesiones cutaneas/lunares que puedan llegar a ser melanoma. La aplicacion, a partir de la entrada de una imagen de una lesion en la piel que tiene potencial de ser peligros y/o dañina, realiza una caracterizacion geometrica de la lesion, una extraccion de caracteristica y finalmente una clasificacion. La caracterizacion geometrica se realiza mediante algoritmos de segmentacion basados en umbralizacion global. La extraccion de caracteristicas se realiza mediante un algoritmo de calculo de la dimension fractal llamado CountingBox. La clasificacion es realizada mediante un algoritmo de clasificacion basado en redes neuronales que trae implementado Matlab. Todos estos algoritmos han sido implementados en una Guide en el programa Matlab, cuya eficacia en la clasificacion de lesiones en una biblioteca de 33 imagenes ha sido del 84%.
[EN] The TFG consists of the development of an application in Matlab for the semiautomatic detection of cutaneous / lunar lesions that may become melanoma. The application, from the entry of an image of a lesion on the skin that has the potential to be dangerous and / or harmful, performs a geometric characterization of the lesion, a feature extraction and finally a classification. The geometric characterization is made by segmentation algorithms based on global thresholding. The extraction of features is done by means of a fractal dimension calculation algorithm called CountingBox. The classification is done by means of a classification algorithm based on neural networks that Matlab implements. All these algorithms have been implemented in a Guide in the Matlab program, whose efficiency in classifying lesions in a library of 33 images has been 84%.
[EN] The TFG consists of the development of an application in Matlab for the semiautomatic detection of cutaneous / lunar lesions that may become melanoma. The application, from the entry of an image of a lesion on the skin that has the potential to be dangerous and / or harmful, performs a geometric characterization of the lesion, a feature extraction and finally a classification. The geometric characterization is made by segmentation algorithms based on global thresholding. The extraction of features is done by means of a fractal dimension calculation algorithm called CountingBox. The classification is done by means of a classification algorithm based on neural networks that Matlab implements. All these algorithms have been implemented in a Guide in the Matlab program, whose efficiency in classifying lesions in a library of 33 images has been 84%.