A discrete tomography approach for superresolution micro-CT images: application to bone

In micro-CT imaging, the effective spatial resolution of the reconstructed images is generally limited by X-ray dose restrictions, the detector configuration or the scanning geometry. In this paper, we show that, using prior information on the grey values of the scanned objects, the spatial resolution of the reconstructed images can dramatically be improved. The proposed method is based on an upsampling of the reconstruction grid, combined with the DART algorithm (discrete algebraic reconstruction technique [1]), in which the scanned object is assumed to be composed of homogeneous materials. Experiments were run on simulated data as well as real X-ray CT data of the rat trabecular bone. Results show that the proposed method generates reconstructions with significantly more detail compared to conventional reconstruction algorithms.