by Shiras Abdurahman, Robert Frysch, Richard Bismark, Steffen Melnik, Oliver Beuing, Georg Rose
Abstract:
The polychromatic X-ray spectrum and the energydependent attenuation coefficient of materials cause beam hardening artifacts in CT reconstructed volumes. These artifacts appear as cupping and streak artifacts depending on the material composition and the geometry of the imaged object. CT scanners employ projection linearization to transform polychromatic attenuation to monochromatic attenuation using a polynomial model. Polynomial coefficients are computed during calibration or using prior information such as X-ray spectrum and attenuation properties of the materials. In this paper, we are presenting a novel method to correct beam hardening artifacts by enforcing cone beam consistency conditions on the projection data. We used consistency conditions derived from Grangeat's fundamental relation between cone beam projection data and 3D Radon transform. The optimal polynomial coefficients for artifact reduction are iteratively estimated by minimizing the inconsistency of a set of projection pairs. The results from simulated and real datasets show the visible reduction of artifacts. Our studies also demonstrate the robustness of the algorithm when the projections are perturbed with other physical measurement and geometrical errors. The proposed method requires neither calibration nor prior information like X-ray spectrum, attenuation properties of the materials and detector response. The algorithm can be used for beam hardening correction in clinical, pre-clinical and industrial CT systems.
Reference:
Beam hardening correction using cone beam consistency conditions (Shiras Abdurahman, Robert Frysch, Richard Bismark, Steffen Melnik, Oliver Beuing, Georg Rose), In IEEE Transactions on Medical Imaging, 2018.
Bibtex Entry:
@article{abdurahman_beam_2018,
title = {Beam hardening correction using cone beam consistency conditions},
issn = {0278-0062},
doi = {10.1109/TMI.2018.2840343},
abstract = {The polychromatic X-ray spectrum and the energydependent attenuation coefficient of materials cause beam hardening artifacts in CT reconstructed volumes. These artifacts appear as cupping and streak artifacts depending on the material composition and the geometry of the imaged object. CT scanners employ projection linearization to transform polychromatic attenuation to monochromatic attenuation using a polynomial model. Polynomial coefficients are computed during calibration or using prior information such as X-ray spectrum and attenuation properties of the materials. In this paper, we are presenting a novel method to correct beam hardening artifacts by enforcing cone beam consistency conditions on the projection data. We used consistency conditions derived from Grangeat's fundamental relation between cone beam projection data and 3D Radon transform. The optimal polynomial coefficients for artifact reduction are iteratively estimated by minimizing the inconsistency of a set of projection pairs. The results from simulated and real datasets show the visible reduction of artifacts. Our studies also demonstrate the robustness of the algorithm when the projections are perturbed with other physical measurement and geometrical errors. The proposed method requires neither calibration nor prior information like X-ray spectrum, attenuation properties of the materials and detector response. The algorithm can be used for beam hardening correction in clinical, pre-clinical and industrial CT systems.},
journal = {IEEE Transactions on Medical Imaging},
author = {Abdurahman, Shiras and Frysch, Robert and Bismark, Richard and Melnik, Steffen and Beuing, Oliver and Rose, Georg},
year = {2018},
keywords = {Attenuation, Beam hardening correction, Bones, Calibration, computed tomography, Cone beam computed tomography, Consistency conditions, Detectors, Grangeat’s fundamental relation, image reconstruction, X-ray imaging},
pages = {1--1}
}