Beam-hardening corrections through a polychromatic projection model integrated to an iterative reconstruction algorithm. (March 2022)

• Record Type: Journal Article
• Authors: Di Schiavi Trotta, Leonardo; Matenine, Dmitri; Martini, Margherita; Stierstorfer, Karl; Lemaréchal, Yannick; Francus, Pierre; Després, Philippe
• Is Part Of: NDT & E international. Volume 126(2022)
• Publication Date: 2022-03
• Subjects: Beam-hardening correction -- Iterative reconstruction -- Computed tomography -- GPU acceleration -- Dual-energy computed tomography; Nondestructive testing -- Periodicals; Contrôle non destructif -- Périodiques; Electronic journals; 620.1127
• DOI: 10.1016/j.ndteint.2021.102594
• Languages: English
• ISSNs: 0963-8695

Abstract

Abstract: In this work, a novel physics-rich beam-hardening correction algorithm was developed for X-ray Computed Tomography. This method uses the spectrum information, the detector response, the filter geometry and a calibration curve. The correction, which does not require prior material knowledge, was embedded in an iterative reconstruction algorithm, and simulates the beam-hardening by estimating the X-ray spectrum at each voxel in the forward projection step. As a result, beam-hardening artifacts are inherently reduced in reconstructed images. For a regular reconstruction matrix of 512 pixels × 512 pixels, processing times of approximately ∼5 s per slice were obtained using a four GPU setup. This method was also compared to the dual-energy beam-hardening correction method proposed by Alvarez and Macovski, which it outperforms when high-Z elements are involved. Highlights: Polychromatic projection model reduces beam-hardening artifacts in X-ray CT. Image artifacts are reduced using CT scanner information: filter and X-ray spectrum. Numerically intense CT reconstructions with X-ray spectrum are accelerated by GPUs.