Adaptive filtering method for reducing signal saturation induced artifacts in X-ray dark-field tomography

Abstract

Abstract X-ray dark-field imaging provides information of an object based on its small-angle scattering properties. Dark-field contrast originates from differences in the strength of scattering from micron and sub-micron sized structures in the object. A Talbot-Lau interferometer can be utilized for dark-field imaging with regular X-ray tubes. A known difficulty with the method arises, when the scattering is too strong, preventing the dark-field and phase retrieval to work accurately. This is similar to a strongly absorbing object in attenuation-based X-ray imaging, and causes problems in tomographic reconstruction that depends on the accuracy of the projected signal. In this article, we present an adaptive filtering approach for the dark-field and phase-contrast projections to improve the signal-to-noise ratio in image areas where the dark-field signal is close to saturation. This filtering improves the projection images markedly, and enables tomographic reconstruction even in cases where the sample contains strong scatterers.