Use this link to cite:
http://hdl.handle.net/2183/39620 Segmentación de tecidos en microscopía histopatolóxica mediante redes neuronais completamente convolucionais
Loading...
Identifiers
Publication date
Authors
Bouzas Quiroga, Jacobo
Advisors
Other responsabilities
Universidade da Coruña. Facultade de Informática
Journal Title
Bibliographic citation
Type of academic work
Academic degree
Abstract
[Resumo]: Explórase a aplicación de técnicas de aprendizaxe automática ao ámbito da histopatoloxía, centrándose na segmentación semántica de imaxes de tecidos. A investigación céntrase no desenvolvemento e adestramento de modelos baseados en redes neuronais convolucionais, co obxectivo de automatizar e mellorar a precisión na diagnose patolóxica. A implantación de arquitecturas avanzadas, como a U-Net, combinada con técnicas de acrecentamento de datos, demostrou ser prometedora para capturar detalles morfolóxicos complexos e mellorar a capacidade de xeneralización dos modelos. Os resultados iniciais obtidos tras o desenvolvemento do proxecto son satisfactorios. Con todo, o traballo permitíu identificar varias direccións de traballo futuro que permitan melloralos, incluíndo a exploración de novos modelos, a implantación de técnicas de aprendizaxe transferida e a avaliación do rendemento con datos adicionais.
[Abstract]: We explore the application of machine learning techniques in the field of histopathology, focusing on semantic image segmentation of tissue samples. The research centers on the development and training of convolutional neural network-based models to automate and enhance diagnostic accuracy. The implementation of advanced architectures, such as U-Net, combined with data augmentation techniques, has proven promising in capturing complex morphological details and improving model generalization. The initial results achieved after the project’s development are satisfactoy. Moreover, the work provided insights into several directions for future work that would allow improvement, including the exploration of alternative models, the application of transfer learning techniques, and performance evaluation with additional data.
[Abstract]: We explore the application of machine learning techniques in the field of histopathology, focusing on semantic image segmentation of tissue samples. The research centers on the development and training of convolutional neural network-based models to automate and enhance diagnostic accuracy. The implementation of advanced architectures, such as U-Net, combined with data augmentation techniques, has proven promising in capturing complex morphological details and improving model generalization. The initial results achieved after the project’s development are satisfactoy. Moreover, the work provided insights into several directions for future work that would allow improvement, including the exploration of alternative models, the application of transfer learning techniques, and performance evaluation with additional data.
Description
Keywords
Redes neuronais completamente convolucionais Análise de imaxes histolóxicas Patoloxía dixital Aprendizaxe profunda Histology image analysis Digital pathology Deep learning Semantic segmentation Breast cancer Computer vision Fully convolutional neural networks Segmentación semántica Cancro de mama Visión artificial
Editor version
Rights
Atribución-NoComercial-CompartirIgual 3.0 España








