Edge-Localized Iterative Reconstruction for Computed Tomography

Recently it has been shown that model-based reconstruction (MBR) can greatly improve the quality of computed tomography (CT) images. In particular, MBR can recover fine details and small features in the reconstruction more accurately than conventional algorithms. In order to fully benefit from this higher spatial resolution, MBR reconstruction requires a higher spatial sampling rate, or equivalently smaller voxels, to represent fine details such as edges. However, these higher spatial sampling rates generate many more voxels for a fixed region-of-interest, so the resulting computation required for reconstruction can be greatly increased. In this paper, we introduce an edge-localized iterative reconstruction algorithm that produces high resolution images at a fraction of the computational cost associated with the conventional full update method. The proposed algorithm works by focusing computation only on the regions of the image that contain fine details, such as edges. Experimental results demonstrate that the proposed algorithm can achieve the same visual quality as the full high resolution reconstruction algorithm at significantly reduced computational cost.