Studies in phase contrast x-ray imaging of biological interfaces
thesisposted on 31.01.2017, 04:12 authored by Morgan, Kaye Susannah
This thesis studies the imaging of interfaces using phase contrast x-ray imaging (PCXI) with a synchrotron source, in theory, practice and application. An emphasis on biological interfaces means that the focus is on single-exposure methods of imaging, as desired when imaging a live sample. Firstly, we look at modelling the formation of propagation-based phase contrast fringes from a curved interface, studying the validity of the projection approximation, often used in this process. The emphasis then moves to how such fringes may be best realised in a non-ideal imaging set-up, looking at the coherence provided by the synchrotron source, notably in the presence and absence of a spinning phase diffuser. The effect that the set-up and associated coherence will have on the visibility and detail of phase contrast fringes from an interface is considered. These findings are then related to the detection of subtle biological interfaces, progressing towards the application of phase contrast imaging to the observation of changes in airway health in response to new treatments for cystic fibrosis. The difficulty in resolving these subtle airway interfaces motivates the final section of work; a new single-exposure sensitive method of phase imaging for live samples using a reference grating. The first section of this work looks at a method of simulating the propagation of an x-ray wavefield through a sample and subsequently through free space in order to produce propagation- based phase contrast fringes. This approach uses the projection approximation to describe the effect of the sample on the incident wavefield, an approximation which does not describe the interference or diffraction of rays within the sample volume. It is seen, through comparison with experiment, that for a curved biological interface, diffraction within the sample volume becomes significant for propagation distances comparable to the sample volume dimension. Hence phase contrast fringes from a rounded interface (a good model for many biological interfaces) are underestimated for very short propagation distances. The intensity fringes seen from a curved edge are studied both in the complex Argand plane and through intensity profiles. At longer propagation distances, such as those typically used in biomedical imaging applications, where free-space propagation within the sample volume is a small percentage of the total propagation, the simulation model is found to match experiment. This assertion is verified when the simulation model is compared to an exact solution to the inhomogeneous Helmholtz equation, both in the Argand plane and through intensity profiles. It is also seen that this exact solution for a plane wave incident on a cylinder will simplify to the projection approximation at the centre of the cylinder, where diffraction effects are negligible. This work demonstrates the value of the projection approximation simulation model, which is sufficiently accurate for the propagation distances commonly utilised for phase contrast images, with computational overheads many orders of magnitude less than the exact solution. The effect of a limited transverse coherence width upon these ideal fringes is then considered. Transverse coherence measurements are taken at beamline 20XU of the SPring-8 synchrotron using a prism interferometer set-up. Measurements are taken at both the upstream and downstream hutches for three sizes of beam-defining aperture. More particularly, measurements are taken with and without a spinning diffuser, an apparatus used to create a more uniform field of incident x-rays and remove unwanted phase effects introduced by upstream optics. It is found that while the presence of a diffuser did not decrease the transverse coherence width, the observed magnitude of the complex degree of coherence is decreased. Computer simulations of the imaging set-up show this same decrease in interference fringe visibility from a diffused beam. The simulation model is then extended to look at the effect of a decreased degree of coherence, due to a diffuser, upon phase contrast fringes from an interface. The suppression of detail and visibility of the interface fringe in the presence of a diffuser is observed in both simulation and in experiment. However, both sets of results also show that simply moving the diffuser closer to the source is sufficient to significantly decrease this effect. This is one of several modifications made to the experimental set-up while aiming to resolve subtle airway interfaces in the application work. This application seeks to measure changes in the airway health in response to new treatments for cystic fibrosis, which are tested (and hence imaged) in mouse models of the disease. One such measure of health is the depth of the airway surface liquid (ASL), which lines the inside of the airways and enables inhaled debris to be cleared from the lungs. While the interface between the ASL and the air-filled inside of the airway produces a strong propagation-based PCXI fringe set, the interface between the ASL and surrounding tissue provides a much weaker phase gradient which is not so easily detected. We detail experimental and analytical methods for increasing the visibility of fringes from this interface and ways to extract information about the ASL depth. Finally, we describe and demonstrate a new single-exposure, large-grating method for detecting both high and low phase gradients. This is achieved by looking at the downstream distortion of a high visibility reference pattern that is incident upon the sample. The resulting quantitative phase maps from images of a phantom suggest this method may be of value in detecting the weak phase gradient at the ASL/tissue interface.