Advances in X-ray reconstruction algorithms for limited data problems in conventional and non-conventional projection geometries

Date: 29 August 2018

Venue: Campus Groenenborger, U0.24 - Groenenborgerlaan 171 - 2020 Antwerpen (route: UAntwerpen, Campus Groenenborger)

Time: 4:00 PM

Organization / co-organization: Department of Physics

PhD candidate: Eline Janssens

Principal investigator: Jan Sijbers & Jan De Beenhouwer

Short description: PhD defence Eline Janssens - Faculty of Science, Department of Physics


X-ray Computed Tomography is a very powerful imaging technique that allows to visualize the internal structure of an object non-destructively. Phase Contrast Computed Tomography is an extension to X-ray Computed Tomography with which both the distribution of the whole imaginary refractive index and the scattering inside an object can be visualized. For both imaging techniques, the quality of a reconstruction is highly dependent on the number of X-ray projections and their angular distribution. Often, only limited projection data can be acquired, for which conventional reconstruction algorithms fail to provide adequate reconstructions.

Moreover, non-conventional acquisition geometries impose extra challenges on the reconstruction process. The central theme throughout this thesis is the search for reconstruction algorithms that still provide adequate reconstructions in case of limited projection data. The key here is to enrich the algorithm with prior knowledge on the objects that are inspected. The thesis is divided into two parts. The first part focuses on inline inspection of objects with transmission X-ray Computed Tomography. The developed algorithm, the NN-hFBP, is engineered for two different applications: inline inspection of agricultural products and inline inspection of ultrasonically welded parts.

In the second part of the thesis, the main imaging method is Phase Contrast Computed Tomography. Here, an alternative to the Eulerian cradle is proposed to visualize scattering caused by fibers and image fusion and segmentation are used to improve the reconstruction quality of highly corrupted data.