Machine learning models for automatic segmentation of brain lesion in MRI images

Show simple item record

dc.contributor Moyà Alcover, Gabriel
dc.contributor Varona Gómez, Javier
dc.contributor Pineda-Pardo, Jose Ángel
dc.contributor.author Balle Sánchez, Marcos
dc.date 2021
dc.date.accessioned 2022-03-03T08:50:34Z
dc.date.issued 2021-10-11
dc.identifier.uri http://hdl.handle.net/11201/158098
dc.description.abstract [spa] En este trabajo se pretende explorar distintos modelos de aprendizaje automático para segmentar una lesión en el cerebro causada por el tratamiento HIFU a pacientes con Temblor Esencial. Para ello, primero se extraen diferentes conjuntos de características para experimentar con algoritmos de clustering, que son el Gaussian Mixture Model (GMM) y KMeans, y de aprendizaje supervisado, que son las Support Vector Machines (SVM) y Random Forest (RF). Sin embargo, ninguno de estos algoritmos ofrece buenos resultados. Posteriormente se entrenan dos redes neuronales convolucionales, la UNet y la VNet. La principal diferencia entre ambas es que la primera procesa imágenes en dos dimensiones mientras que la segunda lo hace en tres dimensiones. En esta ocasión los resultados mejoran considerablemente. Entre ambas redes, es la VNet la que arroja resultados más prometedores. ca
dc.description.abstract [eng] This paper aims to explore different machine learning models for segmenting a brain lesion caused by HIFU treatment of patients with Essential Tremor. To do so, we first extract different feature sets to experiment with clustering algorithms (Gaussian Mixture Model and K-Means) and supervised learning algorithms (Support Vector Machines and Random Forest). However, none of these algorithms gives good results. Subsequently, two convolutional neural networks, UNet and VNet, are trained. The main difference between the two is that the former processes images in two dimensions while the latter processes images in three dimensions. On this occasion, the results are considerably better. Between the two networks, it is the VNet that shows the most promising results. ca
dc.format application/pdf
dc.language.iso spa ca
dc.publisher Universitat de les Illes Balears
dc.rights all rights reserved
dc.rights info:eu-repo/semantics/openAccess
dc.subject 616.89 - Psiquiatria. Psicopatologia ca
dc.title Machine learning models for automatic segmentation of brain lesion in MRI images ca
dc.type info:eu-repo/semantics/masterThesis ca
dc.type info:eu-repo/semantics/publishedVersion
dc.date.updated 2022-02-01T07:09:46Z
dc.date.embargoEndDate info:eu-repo/date/embargoEnd/2050-01-01
dc.embargo 2050-01-01
dc.rights.accessRights info:eu-repo/semantics/embargoedAccess


Files in this item

This item appears in the following Collection(s)

Show simple item record

Search Repository


Advanced Search

Browse

My Account

Statistics