A machine learning based method for sensitivity estimation for accelerated magnetic resonance spectroscopy imaging using phased array coils
Perkins, Kevin
Content Files
PERKINS-THESIS-2018.pdf
Loading…
Download Files
Loading…
Download Counts (All Files)
Loading…
Edit File
Loading…
Permalink
https://hdl.handle.net/2142/100976
Description
Title
A machine learning based method for sensitivity estimation for accelerated magnetic resonance spectroscopy imaging using phased array coils
Author(s)
Perkins, Kevin
Issue Date
2018-04-15
Director of Research (if dissertation) or Advisor (if thesis)
Liang, Zhi-Pei
Department of Study
Electrical & Computer Eng
Discipline
Electrical & Computer Engr
Degree Granting Institution
University of Illinois at Urbana-Champaign
Degree Name
M.S.
Degree Level
Thesis
Date of Ingest
2018-09-04T20:27:05Z
Keyword(s)
Magnetic Resonance Spectroscopic Imaging, MRSI, Parallel Imaging, SENSE, Deep Learning, DnCNN
Abstract
Magnetic resonance spectroscopic imaging (MRSI) enables in-vivo analysis of the spatial distribution of chemicals within the human body. Through MRSI, one can infer the concentration of various metabolites in different regions throughout the body. While the medical implications of such an imaging paradigm are remarkable, a poor trade-off between imaging speed and image resolution has stunted development of MRSI applications.
A combination of many technological advancements is necessary to bring MRSI to its full potential; one advancement is an accelerated imaging technique known as parallel imaging. Parallel imaging exploits differences in receiver sensitivities in phased array coils to recover additional location information. Accurate estimation of the sensitivity profiles is necessary to prevent parallel imaging induced artifacts. However, accurate sensitivity profile estimations require fully sampled high-resolution images which adds an excessive data acquisition burden.
A novel sensitivity profile estimation strategy which relies on deep learning is presented. It is shown how prior information in the form of learned image feature representations may be combined with noisy imaging data to produce high-resolution, artifact-free sensitivity profiles. An in-vivo experiment demonstrates the effectiveness of the proposed method. The relative SENSE reconstruction error for the proposed method is 1.96% compared to a signal processing baseline of 2.52%.
Use this login method if you
don't
have an
@illinois.edu
email address.
(Oops, I do have one)
IDEALS migrated to a new platform on June 23, 2022. If you created
your account prior to this date, you will have to reset your password
using the forgot-password link below.
PERKINS-THESIS-2018.pdf