X-ray CT scatter correction by a physics-motivated deep neural network
Iskender, Berk
Loading…
Permalink
https://hdl.handle.net/2142/109445
Description
Title
X-ray CT scatter correction by a physics-motivated deep neural network
Author(s)
Iskender, Berk
Issue Date
2020-12-08
Director of Research (if dissertation) or Advisor (if thesis)
Bresler, Yoram
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
Keyword(s)
Computed Tomography (CT)
Deep learning
X-ray CT scatter
Scatter estimation
Scatter correction
Physics-motivated
Monte-Carlo
Abstract
A fundamental problem in X-ray Computed Tomography (CT) is the scatter occurring due to the interaction of photons with the imaged object. Unless it is corrected, this phenomenon manifests itself as degradations in the reconstructions in the form of various artifacts. This makes scatter correction a critical step to obtain the desired reconstruction quality.
Scatter correction methods consist of two groups: hardware-based and software-based. Despite success in specific settings, hardware-based methods require modification in the hardware or an increase in the scan time or dose. This makes software-based methods attractive. In this context, Monte-Carlo based scatter estimation, analytical-numerical and kernel-based methods were developed. Furthermore, the capacity of data-driven approaches to tackle this problem was recently demonstrated.
In this thesis, two novel physics-motivated deep-learning-based methods are proposed. The methods estimate and correct for the scatter in the obtained projection measurements. They incorporate both an initial reconstruction of the object of interest and the scatter-corrupted measurements related to it. They use a common specific deep neural network architecture and a cost function adapted to the problem.
Numerical experiments with data obtained by Monte-Carlo simulations of the imaging of phantoms reveal noticeable improvement over a recent projection-domain deep neural network correction method.
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.
