Cone-beam Reconstruction for Micro-CT

X-ray micro-CT has been a focus in the CT/Micro-CT Lab, the University of Iowa. In this paper, some recent progress we made on cone-beam micro-CT is described. First, we address approximation reconstruction using Feldkamp-type algorithms with multiple X-ray sources and displaced detector configurations. After an appropriate fan-beam weighting function on the rectangular redundant region for equi-angular data is derived for multiple imaging chains, we can insert it into a general Feldkamp algorithm for cone-beam reconstruction with multiple sources and obtain a reconstruction formula. Then, we explain three types of artifacts associated with exact reconstruction in the Grangeat framework as applied to the circular scanning locus, which are thorn, wrinkle, and V artifacts. The thorn pattern is due to inappropriate extrapolation into the shadow zone in the Radon domain. Hence, the zero padding technique should be avoided in this context. The wrinkle texture arises if interpolation needed to compute the first derivatives of the Radon data is not smooth between adjacent detector planes. In particular, the nearest neighbor interpolation method should not be used in general. If the number of projections is not small, the bi-linear interpolation method is effective to suppress the wrinkle artifacts. The V shape on the meridian plane comes from the line integrations through the transition zones where derivative data change abruptly, and are very unstable. Two immediate remedies are to increase the sampling rate and suppress data noise. Finally, we report our recent convergence results of a generalized block-iterative Landweber scheme, which includes the SART and some other well-known algorithms and their ordered-subset versions as special cases. The iterative approach has been important in image reconstruction for reducing image noise and artifacts in the cases of noisy and/or incomplete data. It is shown that block-iterative schemes can greatly speed the reconstruction process and produce satisfactory image. Representative images from simulation and experiments are also given.