A Multi-resolution Fast Reconstruction Algorithm for X-Ray In-Line Phase Contrast CT with Sparse Sampling

X-ray in-line phase contrast computed tomography (PC-CT) with sparse sampling is one of the novel non-destructive imaging methods and has been popular in medicine, biomedical engineering and material sciences. It usually adopts iterative algorithm with total variance theory to reconstruction image. Due to the existence of iteration operation, the reconstruction is time-consuming. In this paper, a multi-resolution reconstruction algorithm is proposed to implement the fast image reconstruction for in-line PC-CT with sparse sampling. The use of resolution stepped-up is the most iconic design of the new method. In this method, the low resolution image is first reconstructed quickly with less data and smaller discrete imaging model. Adopting the low resolution image as the initial image, high resolution iterative reconstruction is then executed with the whole data. The simulation results with the well-known Shepp-Logan three dimensional head phantom demonstrate the validity of the proposed method. The reconstruction time was 50% smaller than that of the conventional method for the phantom with a size of 400*400*400 pixels. It can also save computer memory.