Abstract

Info

Title:: Lung 4D-CT image super-resolution reconstruction using self-similarity-based multiscale sparse representation

Author(s):: WANG Tingting1; ZHANG Yu2; 1. Department of Medical Equipment, Nanfang Hospital, Southern Medical University, Guangzhou 510515, China; 2. School of Biomedical Engineering, Southern Medical University, Guangzhou 510515, China

Keywords:: ; Keywords: lung; four-dimensional computed tomography; super-resolution reconstruction; image self-similarity; multiscale analysis; sparse representation

Abstract:: Abstract: Based on the anisotropy that the interlayer resolution of lung four-dimensional computed tomography (4D-CT) image is far lower than the intralayer resolution, a self-similarity-based multiscale sparse representation for super-resolution reconstruction is developed to enhance the resolution of lung 4D-CT image. The self-similarity of transverse and coronal/sagittal lung 4D-CT images is explored with the proposed approach. Based on the validity of cross-scale self-similarity, the transverse images of lung 4D-CT data instead of coronal/sagittal images are utilized as training database for learning the mapping between low-resolution and high-resolution image patches. Subsequently, multiscale strategy is introduced. The image patches of different scales are obtained based on quad-tree decomposition. After training, multiscale global dictionaries are obtained for capturing more anatomical features. Finally, the coronal/sagittal high-resolution images are reconstructed by sparse representation based super-resolution algorithm. Herein the simulated and real data are employed to verity the effectiveness of the proposed approach. Both quantitative evaluation and visual results reveal that the proposed approach has good performances in generating clear image with significantly enhanced structures, and meanwhile effectively avoids the process of motion evaluation which restricts the efficiency and accuracy of reconstructed result.