Time-resolved X-ray phase-contrast tomography

by Aike Ruhlandt
Doctoral thesis
Date of Examination:2017-11-27
Date of issue:2018-06-27
Advisor:Prof. Dr. Tim Salditt
Referee:Prof. Dr. Tim Salditt
Referee:Prof. Dr. Jörg Enderlein
Referee:Prof. Dr. Claus Ropers
Referee:Dr. Alexander Egner
Referee:Prof. Dr. Sarah Köster
Referee:Prof. Dr. Reiner Kree
Persistent Address: http://dx.doi.org/10.53846/goediss-6754

 

 

Files in this item

Name:dissA4_linearized.pdf
Size:48.7Mb
Format:PDF
Description:Dissertation Aike Ruhlandt 2017

The following license files are associated with this item:

Abstract

English

X-ray tomography allows to obtain the three-dimensional (3d) structure of weakly interacting objects, such as tissue samples in biomedical imaging or nanoscale structures in materials science, with high spatial resolution. However, sufficient contrast can often only be achieved when employing phase contrast techniques. One of the latter is in-line holography, a scanning-free method that, in combination with highly brilliant synchrotron sources, enables rapid image acquisition. This opens up the possibility to study 3d dynamics in time-resolved X-ray phase-contrast tomography. This work first presents methods for fast and high quality numerical propagation and approaches for combined reconstruction of the phase-information and the 3d structure from a set of holograms. In particular, it is shown how tomographic consistency can be used as a powerful constraint in phase-retrieval, making use of the novel concept of virtual free-space propagation of entire 3d objects. Attention is brought to several practical aspects concerning the experimental realisation of time-resolved tomography, as well as the numerical implementation of the data analysis steps. Two types of samples -- namely a burning wooden match and sedimenting silica micro-spheres in a water-filled capillary -- are used to demonstrate some of the challenges of time-resolved phase-contrast tomography, and how they can be overcome. In this context, a new approach to minimize motion-related artefacts in tomography is presented. The technique, termed ``backpropagation along dynamically curved paths", is based on the development of an initial motion model using an optical flow analysis of direct reconstructions. The motion model is then fed into the tomographic reconstruction geometry, allowing to reconstruct structure and dynamics simultaneously with considerably improved quality. Second, the trajectories of thousands of sedimenting micro-spheres are measured simultaneously. The reconstruction provides a detailed insight into the processes and hydrodynamic interactions in a sedimentation regime that is barely accessible by established methods.
Keywords: x-ray; tomography; phase-retrieval; propagation; synchrotron; sedimentation; burning match; dynamic
 
Statistik