Dictionary learning based sinogram inpainting for CT sparse reconstruction |
| |
Authors: | Si Li Qing Cao Yang Chen Yining Hu Limin Luo Christine Toumoulin |
| |
Affiliation: | 1. Laboratory of Image Science and Technology, Southeast University, Nanjing, China;2. Centre de Recherche en Information Biomedicale Sino-Francais (LIA CRIBs) , Rennes, France;3. Laboratoire Traitement du Signal & de l’Image (LTSI)-Université de Rennes, F-35042 Rennes, France;4. INSERM U1099, Université de Rennes, F-35042 Rennes, France |
| |
Abstract: | In CT (computed tomography), reconstruction from undersampling projection data is often ill-posed and suffers from severe artifact in the reconstructed images. To overcome this problem, this paper proposes a sinogram inpainting method based on recently rising sparse representation technology. In this approach, a dictionary learning based inpainting is used to estimate the missing projection data. The final image can be reconstructed by the analytic filtered back projection (FBP) reconstruction. We conduct experiments using both simulated and real phantom data. Compared to the comparative interpolation method, visual and numerical results validate the clinical potential of the proposed method. |
| |
Keywords: | Computed tomography (CT) Under-sampling Dictionary learning Inpainting |
本文献已被 ScienceDirect 等数据库收录! |
|