Backprojection Wiener deconvolution for computed tomographic reconstruction

Zhenglin Wang; Jinhai Cai; William Guo; Martin Donnelley; David Parsons; Ivan Lee

doi:10.1371/journal.pone.0207907

Backprojection Wiener deconvolution for computed tomographic reconstruction

PLoS One. 2018 Dec 18;13(12):e0207907. doi: 10.1371/journal.pone.0207907. eCollection 2018.

Authors

Zhenglin Wang¹, Jinhai Cai², William Guo¹, Martin Donnelley^{3

4}, David Parsons^{3

4}, Ivan Lee²

Affiliations

¹ Centre for Intelligent Systems, School of Engineering and Technology, Central Queensland University, North Rockhampton, QLD, Australia.
² School of Information Technology and Mathematical Sciences, The University of South Australia, Mawson Lakes, SA, Australia.
³ Respiratory and Sleep Medicine, Women's and Children's Hospital, North Adelaide, SA, Australia.
⁴ Robinson Research Institute and Adelaide Medical School, University of Adelaide, North Adelaide, SA, Australia.

Abstract

Analytical CT reconstruction is popular in practice because of its computational efficiency, but it suffers from low reconstruction quality when an insufficient number of projections are used. To address this issue, this paper presents a new analytical method of backprojection Wiener deconvolution (BPWD). BPWD executes backprojection first, and then applies a Wiener deconvolution to the whole backprojected image. The Wiener filter is derived from a ramp filter, enabling the proposed approach to perform reconstruction and denoising simultaneously. The use of a filter after backprojection does not differentiate between real sampled projections and interpolated ones, introducing reconstruction errors. Therefore a weighted ramp filter was applied to increase the contribution of real sampled projections in the reconstruction, thus improving reconstruction quality. Experiments on synthetic data and real phase-contrast x-ray images showed that the proposed approach yields better reconstruction quality compared to the classical filtered backprojection (FBP) method, with comparable reconstruction speed.

Publication types

Research Support, Non-U.S. Gov't

MeSH terms

Algorithms
Animals
Computer Simulation
Humans
Lung / diagnostic imaging
Mice
Models, Statistical
Phantoms, Imaging
Radiographic Image Interpretation, Computer-Assisted / statistics & numerical data*
Signal-To-Noise Ratio
Tomography, X-Ray Computed / statistics & numerical data*

Grants and funding

This work was supported by Imaging and Medical Beamline at the Australian Synchrotron, Victoria, Australia under proposals M5211 and M7028 (http://www.synchrotron.org.au/, role: data collection), Women’s & Children’s Hospital Foundation (https://www.wchfoundation.org.au/default.aspx, role: The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript), Cure 4 Cystic Fibrosis Foundation (http://www.cure4cf.org/, role: The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript), and National Health and Medical Research Council project GNT1079712 (https://www.nhmrc.gov.au/, role: The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript). MD was supported by a Robinson Research Institute Career Development Fellowship (role: The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript). ZW was supported by a Central Queensland University Early Career Fellowship (role: The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript).