Santos, Olga C.Manzano Patron, Jose PedroRapado Morales, Dayris2025-10-102025-10-102024-06-01Rapado Morales, Dayris. Trabajo Fin de Máster: Exploring self-supervised learning denoising in diffusion MRI - A characterisation of their impact under different noise regimes. Universidad Nacional de Educación a Distancia (UNED), 2024https://hdl.handle.net/20.500.14468/30362Diffusion Magnetic Resonance Imaging (dMRI) is the main non-invasive technique for understanding brain structural connectivity. However, given its inherent low signal-to-noise ratio, thermal noise particularly affects this modality and its downstream analysis. To overcome this challenge, several denoising methods have been developed over the years. Based on previous works, and as an extension, we assess and characterise here the impact of some of the selfsupervised Deep Learning approaches recently developed for denoising and show comparisons with alternative traditional approaches in both synthetic and real brain datasets.eninfo:eu-repo/semantics/openAccess33 Ciencias Tecnológicastesis de maestríaDiffusion MRIdenoisingself-supervised learningdiffusion modelsGenerative modelsMPPCANLMPatch2SelfDDM2Noise flooruncertainty