Removing multiple outliers and single-crystal artefacts from X-ray diffraction computed tomography data

This paper reports a simple but effective filtering approach to deal with single-crystal artefacts in X-ray diffraction computed tomography (XRD-CT). In XRD-CT, large crystallites can produce spots on top of the powder diffraction rings, which, after azimuthal integration and tomographic reconstruction, lead to line/streak artefacts in the tomograms. In the simple approach presented here, the polar transform is taken of collected two-dimensional diffraction patterns followed by directional median/mean filtering prior to integration. Reconstruction of one-dimensional diffraction projection data sets treated in such a way leads to a very significant improvement in reconstructed image quality for systems that exhibit powder spottiness arising from large crystallites. This approach is not computationally heavy which is an important consideration with big data sets such as is the case with XRD-CT. The method should have application to two-dimensional X-ray diffraction data in general where such spottiness arises.