Summary
IEICE Information and Communication Technology Forum
2017
Session Number:SESSION03
Session:
Number:SESSION03_2
Greedy MRI reconstruction using Markov Random Field prior
Marko Panic, Dejan Vukobratovic, Vladimir Crnojevic, Aleksandra Pizurica,
pp.-
Publication Date:2017/10/1
Online ISSN:2188-5079
DOI:10.34385/proc.50.SESSION03_2
Summary:
Recent work on compressed sensing in magnetic resonance imaging (CS-MRI) indicates benefits of modellingthe structure of sparse coefficients. Comprehensive studies are available for tree-structured models. Much less work has been done on using statistical models for intra-scale (spatial) dependencies, like Markov Random Field (MRF) models in CS-MRI, although initial studies showed great potentials. We present herean efficient greedy algorithm with MRF priors and demonstrate encouraging performance in comparison to related methods, including those based on tree-structured sparsity.