Please use this identifier to cite or link to this item: http://hdl.handle.net/10609/138567
Title: Clasificación de lesiones dermatológicas a partir de imágenes dermoscópicas mediante el aprendizaje automático
Author: Montes Mantero, Francisco Jesús
Tutor: Torre Gallart, Jordi de la
Others: Prados Carrasco, Ferran  
Abstract: Dermatological lesions have a high incidence in the human population. However, its consequences can be mitigated if they are diagnosed early. In this sense, it would be interesting to build tools for healthcare professionals who are capable of helping in its early detection and diagnosis. The objective of this work is to create a model that, using machine learning techniques, develops the ability to recognize and classify these types of lesions correctly from dermoscopic images of patients. A set of dermoscopic images corresponding to 9 categories of well-defined dermatological lesions from the ISIC 2019 archive has been considered. This collection of images, cataloged and labeled by professional dermatologists, will serve as the basis for training a convolutional neural network model that will classify them. In this work, exclusively neural models based on the EfficientNet family of architectures are used. The general method that has been followed consists of the following stages: 1.- Compilation of information on the state of the art of image classification using neural models. 2.- Obtaining candidate EfficientNet models, training them using state-of-the-art techniques and also resorting to assembly techniques. 3.- Evaluation of the candidate models based on the score of the objective measure defined in the ISIC 2019 competition. According to the target measure, an assembled EfficientNet model turned out to be more suitable than the best of its component models. However, this improvement did not turn out to be absolute when more specific metrics were examined.
Keywords: dermatology
diseases
machine learning
Document type: info:eu-repo/semantics/masterThesis
Issue Date: Jan-2022
Publication license: http://creativecommons.org/licenses/by-nc-nd/3.0/es/  
Appears in Collections:Bachelor thesis, research projects, etc.

Files in This Item:
File Description SizeFormat 

uranio255TFM0122vídeo.mp4

Video presentación del TFM142,98 MBMP4View/Open
uranio255TFM0122memoria.pdfMemoria del TFM2,31 MBAdobe PDFThumbnail
View/Open