Generalized Variation Minimization Based Algorithm for Image Reconstruction from Few-Views of Limited Angular in Computed Tomography

Because of the difficulty of image reconstruction from limited angular range with few-views projection data in computed tomography (CT), a generalized variation (GV) minimization based image reconstruction algorithm is developed from few-views in limited angular range in this paper. In the new algorithm, the minimization of the GV norm is used in image reconstruction model as a constrained condition to modify the reconstructed results. To solve the mathematical model efficiently, conjugate gradient (CG) algorithm is applied to solve the minimization problem of the GV norm. Furthermore, CG method is also used in the iteration process to solve the algebraic equation based on the projection geometry in order to improve the convergence rate. We have performed numerical experiments using both computer simulation data and real data in medical CT. Experimental results show that comparison with the traditional CT image reconstruction algorithms such as filter back projection (FBP) and algebraic reconstruction technique (ART), the proposed algorithm can improve the quality of reconstructed results with few views projections data from limited angles.