A Unified Probabilistic Computational Framework for Cross-Domain Compensated Medical Imaging
Loading...
Date
2016-12-23
Authors
Boroomand, Ameneh
Journal Title
Journal ISSN
Volume Title
Publisher
University of Waterloo
Abstract
The quality of medical imaging is an important factor that can directly affect on the results
of different medical imaging tests which routinely being used for various clinical studies
and related research. The existence of any type of image imperfection such as different
noise and artifacts can make some difficulties for the correct analyzing and consequent
interpretation of medical imaging test results and even may impact on the overall decision
of the performed clinical procedure. Due to this issue, there is always a motivation for the
quality enhancement of medical imaging that needs to be done by correcting for the different
image degradation that typically arise during medical image capture or reconstruction.
Computational compensated imaging provides a useful, cheap and easy solution for
the quality enhancement of different types of medical imaging when it compares to the
similar hardware-based compensation methods. A computational compensated medical
imaging basically aims to recover a compensated (true) image from degraded medical
measurement that is affected by the different types of acquisition/reconstruction image
degradation. Compensating for each single image degradation issue can be useful for the
quality improvement of medical imaging. Having a computational compensated imaging
framework to jointly correct for multiple acquisition/reconstruction degradation issues can
improve the functionality of current existing compensated imaging frameworks as they
mostly account only for a single specific type of degradation issue.
This work presents a novel probabilistic based computational compensated medical
imaging which is able to jointly account for several acquisition/reconstruction degradation
issues from the different domains in a unified computational framework. The developed
computational cross-domain compensated medical imaging specifically takes advantage of
a stochastically fully connected conditional random field (SFC-CRF) model in its frame-
work which improves the performance of the proposed compensated medical imaging in
producing of a desired compensated medical image.
A compensated optical coherence tomography (C-OCT) imaging is developed within
the framework of proposed computational cross-domain compensated medical imaging and
with aiming to jointly compensate for the degradation due to the optical aberrations and
speckle noise in OCT imaging. The developed C-OCT imaging is expanded to design
a compensated super resolution OCT (C-SR-OCT) imaging framework which is able to
generate a super resolution OCT (SR-OCT) image of higher quality from multiple OCT
measurements. The proposed computational cross-domain compensated medical imaging is
also used to develop a compensated magnetic resonance imaging (CMR) framework which
aims to improve the quality of MR imaging from different modalities by jointly correcting for the MR image degradation due to the intrinsic properties of MR scanner, bias field
inhomogeneities and inherent MR noise. The results of all three designed compensated
medical imaging platforms for the OCT imaging and MR imaging elaborate the promising efficacy of proposed probabilistic based computational compensated medical imaging
framework for the quality enhancement of different medical imaging techniques.
Description
Keywords
Medical image processing, Medical image quality improvement, Probabilistic modeling