Image Reconstruction using Self-Prior Information for Sparse-View Computed Tomography

Abstract

The contradiction between the great benefits of computed tomography (CT) in diagnosis and the risk of redundant CT scan on the patient health, make the researchers compete developing image reconstruction methods for low-dose CT. Sparse-view CT is a common technique in radiation dose minimization. Due to the streak artifacts that result while using the analytical reconstruction method with sparse-view CT, several iterative reconstruction methods have presented to produce high image quality. In this work, we introduce extracting the prior information incorporated in the reconstruction method during the process of reconstruction itself, in contrast to the other related methods that prepare the prior information in advance. The proposed technique is divided into two main steps. The first step is the construction of self-prior information. The second step is incorporating this produced information into the reconstruction process. The performance of the proposed method is evaluated using simulation and synthetic real data. Results show that the proposed technique produce high image quality.