X-Ray Computed Tomography using Partially Coherent Fresnel Diffraction with Application to an Optical Fiber
Zachary H. Levine, Edward J. Garboczi, Adele P. Peskin, Axel A. Ekman, Elisabeth Mansfield, Jason D. Holm
An algorithm for partially coherent x-ray computed tomography including Fresnel diffraction is developed and applied to an optical fiber. The computing time is a few times worse than the projective counterpart. A code using the algorithm is used to reconstruct a tilt series acquired at the 1~$\mu$m scale of a graded-index optical fiber with projections and diffraction, and using maximum likelihood and a Bayesian method due to Bouman and Sauer. Including Fresnel diffraction removes some reconstruction artifacts. The use of the Bayesian prior removes others.