Reconstruction using multiple images for emerging CT imaging

Emerging computed tomographic (CT) imaging modalities such as low dosage imaging and carbon-nano tube imaging impose new challenges for image reconstruction from projections. The limited signal-to-noise ratio and geometry lead to spatial variant artifacts, making it hard for extracting clinical useful information. We present a novel reconstruction method based on the observation that images reconstructed from different subsets of projections share similar information on object structure but have different information on artifacts. This observation leads to effective artifacts removal and it preserves the edges without sophisticated edge modeling. Experimental results demonstrate the effectiveness of this approach.