Total variation denoising of diffusion MRI images using a modified monge-kantorovich norm
LE3 .A278 2020
Bachelor of Science
Mathematics & Statistics
Efficient methods for denoising various forms of data are contantly evolving and are crucial in fields like medical imaging where even small amounts of noise can affect a doctor’s diagnosis. It has been proposed in a 2016 paper (D. La Torre et al) that using the Monge-Kantorovich metric, along with total variation and regularization, one could create an effective and hopefully efficient method for denoising diffusion MRI images. In this thesis this aforementioned method for denoising will be discussed as well as the advantages of a Monge-Kantorovich style norm versus a Euclidean or other type of norm. Finally the effectiveness of a python-based implementation that was created summer 2019 will be discussed as a measure of the efficiency of the algorithm.
The author retains copyright in this thesis. Any substantial copying or any other actions that exceed fair dealing or other exceptions in the Copyright Act require the permission of the author.