CT Reconstruction from Few-Views by Higher Order Adaptive Weighted Total Variation Minimization

Dose reduction in X-ray Computed Tomography (CT) is of high practical relevance. Compressed Sensing allows for efficient under-sampling while still achieving an acceptable image quality. Especially Total Variation (TV) regularization obtains accurate, robust and stable results. However, it often suffers from the loss of fine structures and stair-casing artifacts. In order to overcome these limitations, we propose a generalization of TV by higher order derivatives. We demonstrate in this paper that both stair-casing and the loss of small structures in TV-based iterative tomographic reconstructions can be overcome.