Compressed Sensing Reconstruction Methods for Quantitative Relaxometry Mapping in MRI: a Comparison

Lead Research Organisation: University College London
Department Name: Medical Physics and Biomedical Eng

Abstract

This MRes project aims to produce a comprehensive comparison of the existing methods of compressed sensing applied to the field of T1/T2 relaxometry in MRI. Compressed sensing (CS) is a new paradigm of acquiring a lower amount of data for MR image formation by exploiting the redundancy of the image information content. While there are many implementations of CS in relaxometry and as many claims of performance over the previous state-of-the-art method, to the best of our knowledge there is no study that reunites diverse methods under a single testing framework. We aim to address this knowledge gap. The goal is to implement different CS algorithms and comparatively assess their strengths and weaknesses. The algorithms will be representative for each of the major approaches in CS: exploiting general sparsity (spatial smoothness, parametric dimension smoothness, low-rank) or using a model-based sparse representation (dictionary, principal components). The comparison involves the essential step of choosing data and relevant metrics. Some potential indicators could be: error of reconstruction against a benchmark, time performance, compatibility with multiple acquisition sequences, applicability to diagnosis.
The MRes project will build the knowledge necessary for developing a novel CS technique applicable to diffusion MRI, which is the envisioned aim of the PhD.

Publications

10 25 50

Studentship Projects

Project Reference Relationship Related To Start End Student Name
EP/R512400/1 01/10/2017 31/03/2022
1921685 Studentship EP/R512400/1 25/09/2017 28/06/2019 Victor Serban