Effect of an initial solution in iterative reconstruction of dynamically changing objects

Visualizing and analyzing dynamic processes in 3D is an emerging topic, e.g. in geosciences (Berg et al., 2009; Cnudde and Boone, 2013; Bultreys et al., accepted), which has only recently become possible due to fast, high-resolution CT scanning. However; dynamically changing objects pose a challenge in CT-imaging because the existing reconstruction algorithms, which reconstruct the sample volume from a number of scan images, presume an unchanging sample during the acquisition of the projection images. Movements or changes during the scan cause artefacts in the resulting volume. Furthermore, when fast processes are visualized, the acquisition time needs to be reduced, thus drastically decreasing the signal-to-noise ratio (SNR). To address these issues, an iterative reconstruction technique is applied, where an initial solution is provided to the algorithm. In this work, we present an evaluation of this method based on both simulations and real experimental data.