X-ray in-line phase contrast micro-tomography is a relatively new coherent X-ray
imaging technique which offers a sensitivity several orders of magnitude higher than
standard, attenuation based techniques. In this technique, phase contrast is achieved
by simply letting the beam propagate after interaction with the imaged object. More
interestingly, the images acquired with this technique can be used to reconstruct the
phase shift induced by the object β phase retrieval. Further, the retrieved phase shifts,
reconstructed at different rotation angles of the object can be used to reconstruct the
3D refractive index distribution in the sample β phase tomography.
Several different phase retrieval algorithms have been presented in literature.
Sensitivity to low frequency noise has required the introduction of priors, such as
assuming a homogeneous object. One of these algorithms reconstructs the phase
from one propagation phase contrast image alone by using such assumptions,
whereas other methods use images at several distances to reconstruct the phase,
while putting fewer restrictions on the object. The obvious gain of a single distance
technique is the simplicity of the experimental setup, reduced amount of recorded
data and reduced use of precious synchrotron beam time, whereas the multi-distance
techniques offers better image quality.
In the multi-distance techniques, the images at different distances have to be aligned,
however. Despite very stable instrumentation, there are always instrumental drifts
and vibrations that must be compensated. The aim of this project is to investigate the
influence of misalignments on the reconstructed image quality in phase imaging and
tomography, both qualitatively and quantitatively. Further, based on the chosen
criteria, different similarity measures and registration algorithms will be evaluated,
for in-line phase contrast images on simulated data and in phase imaging on
experimental data based on the resulting image quality. The chosen algorithm will be
implemented at the European Synchrotron Radiation Facility.
The successful candidate should have a background in computer science, electrical
engineering or engineering physics, and a strong interest in biomedical imaging,
optimization, signal and image processing and related areas. The project will carried
out at the Centre Leon Berard, Lyon.
Laboratory: CREATIS, CNRS UMR 5220, INSERM U1044, UniversitΓ© Lyon 1,
INSA Lyon