Share Email Print
cover

Proceedings Paper

Quantitative low-dose CT: potential value of low signal correction methods
Author(s): Juan Pablo Cruz-Bastida; Ran Zhang; Daniel Gomez-Cardona; Guang-Hong Chen
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

In previous works, it has been demonstrated that CT number estimates are biased at low-dose levels for FBPbased CT. The purpose of this work was to explore the potential of noise-reduction methods to address this limitation. One may speculate that these potential CT number bias issues may have been addressed by noisereduction schemes, such as those implemented in commercial products. However, to the best of the authors’ knowledge, this is not the case. In this work, we performed an in-house noise-reduction implementation on benchtop-CT data, to further investigate under what conditions noise-reduction methods could favor quantitative low-dose CT. A quality assurance phantom was scanned at different dose levels, and the CT number bias was estimated for different material inserts of known composition, prior and after noise reduction. Anisotropic diffusion (AD) filtration was used as noise-reduction method, and independently applied in three different signal domains: raw counts, log-processed sinogram and reconstructed CT image. As result, AD filtration in the raw counts domain was the only noise-reduction scheme successful to considerably reduce CT number bias. Our results suggest that noise-reduction methods can help to preserve CT number accuracy to some degree when reducing radiation dose; however, their application cannot be arbitrarily extended to any signal domain. It was also observed that CT number bias reduction is greater for stronger filtration, which suggest that the overall CT image quality may limit the quantification accuracy when noise-reduction is performed.

Paper Details

Date Published: 7 March 2019
PDF: 6 pages
Proc. SPIE 10948, Medical Imaging 2019: Physics of Medical Imaging, 1094830 (7 March 2019); doi: 10.1117/12.2512889
Show Author Affiliations
Juan Pablo Cruz-Bastida, Univ. of Wisconsin School of Medicine and Public Health (United States)
Ran Zhang, Univ. of Wisconsin School of Medicine and Public Health (United States)
Daniel Gomez-Cardona, Univ. of Wisconsin School of Medicine and Public Health (United States)
Guang-Hong Chen, Univ. of Wisconsin School of Medicine and Public Health (United States)


Published in SPIE Proceedings Vol. 10948:
Medical Imaging 2019: Physics of Medical Imaging
Taly Gilat Schmidt; Guang-Hong Chen; Hilde Bosmans, Editor(s)

© SPIE. Terms of Use
Back to Top