Linear programming-based reconstruction algorithm for limited angular sparse-view tomography

Abstract

Abstract The reconstruction of high-quality images from limited angular sparse-view tomography data is desired in many fields, including low-dose computed tomography and nondestructive detection. In this study, two regularization terms based on the assumptions of image continuity and nonzero pixels are utilized to reconstruct an image from incomplete data. By transforming the original nonlinear reconstruction model into a linear programming, the proposed method can obtain the optimal solution rather than the local minimum point, which helps to accurately reconstruct the images. To demonstrate the effect of the proposed method, we designed a simulation experiment under limited angle, few-view, and limited angular sparse-view conditions, and evaluated the reconstruction results using the structural similarity index, peak signal-to-noise ratio, and cosine distance. The experimental results show that our method can effectively overcome the adverse effects of sparse-view conditions and a limited scanning range. In the tomographic results of our method, streak artifacts do not appear, even when using projection data from 20 views of a 180° scanning angular range.