A method for material decomposition and quantification with grating based phase CT

• Published in: PloS one Vol. 16; no. 1; p. e0245449
• Main Authors: Deng, Shiwo, Zhu, Yining, Zhang, Huitao, Wang, Qian, Zhu, Peiping, Zhang, Kai, Zhang, Peng
• Format: Journal Article
• Language: English
• Published: United States Public Library of Science 22-01-2021; Public Library of Science (PLoS)
• Subjects: Absorption; Absorption coefficient; Absorptivity; Algorithms; Attenuation; Biology and Life Sciences; Bones; Computed tomography; Contrast agents; CT imaging; Decomposition; Decomposition (Chemistry); Drafting software; Editing; Energy; Engineering and Technology; Engineering research; Funding; High energy physics; Image reconstruction; Innovations; Iodine; Iterative methods; Mathematical analysis; Medicine and Health Sciences; Noise; Physical Sciences; Physics; Research and Analysis Methods; Reviews; Scattering coefficient; Technology; X-rays; China; Beijing China
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0245449

Material decomposition (MD) is an important application of computer tomography (CT). For phase contrast imaging, conventional MD methods are categorized into two types with respect to different operation sequences, i.e., "before" or "after" image reconstruction. Both categories come down to two-step methods, which have the problem of noise amplification. In this study, we incorporate both phase and absorption (PA) information into MD process, and correspondingly develop a simultaneous algebraic reconstruction technique (SART). The proposed method is referred to as phase & absorption material decomposition-SART (PAMD-SART). By iteratively solving an optimization problem, material composition and substance quantification are reconstructed directly from absorption and differential phase projections. Comparing with two-step MD, the proposed one-step method is superior in noise suppression and accurate decomposition. Numerical simulations and synchrotron radiation based experiments show that PAMD-SART outperforms the classical MD method (image-based and dual-energy CT iterative method), especially for the quantitative accuracy of material equivalent atomic number.