View Aliasing Artifacts Reduction Method Based on 4D Cone-Beam CT Reconstruction with Joint Projection Data

Shaohua Zhi,Xuanqin Mou
2017-01-01
Abstract:Although the quality of the phase-resolved reconstruction images could be improved by the four-dimensional cone-beam computed tomography (4D-CBCT) by reducing the motion blurring artifacts, it may still be degraded by severe viewaliasing artifacts because of highly under-sampled projections at each phase. Inspired by the strong correlation between different phase-resolved reconstructed images, we present a simple and effective approach to estimate a set of full-sampled projections for every individual respiratory phase and then to incorporate them into the 4D-CBCT iterative reconstruction scheme. In the implementation of the 4D-CBCT iterative reconstruction scheme, a coupled distance-driven forward and backward projection operator via GPU is introduced. The proposed method has been tested in a digital XCAT phantom and a clinical patient case. Quantitative evaluations indicate that a 15.7% and 9.9% decrease in the root-mean-square error (RMSE) are achieved by our method when comparing with the conventional 4D-CBCT reconstruction method and the classic McKinnon/Bates algorithm (MKB), respectively. At the same time, our method is also valid by calculating the contrast-to-noise ratio (CNR) of a region of interest (ROI). The result shows that the CNR of our method is 1.34, which is better than that of the MKB algorithm.
What problem does this paper attempt to address?