Sparse-view CT imaging of trabecular bones: Comparison of image reconstruction methods

Abstract

Purpose In longitudinal in-vivo imaging studies of trabecular bones using a micro-CT, sparse-view imaging is highly desired to reduce x-ray dose to the animals under study and to reduce the scan time. Sparse-view imaging, i.e., imaging with reduced number of views in the CT scan, may result in severe streak artifacts in the reconstructed images degrading the image quality. We compare image reconstruction methods in terms of final image quality in CT imaging of trabecular bones. Methods We have compared three image reconstruction methods, the conventional filtered backprojection method and two iterative image reconstruction methods, i.e., SIRT and TV-minimization methods, in trabecular bone imaging with sparse-view CT scans. After acquiring projection data from sparse-view scanning of a chicken bone using a micro-CT, we have calculated relative image errors and edge preservation metrics from the images reconstructed by the three different methods. Results We have observed that the TV-minimization method outperforms the other methods in trabecular bone imaging in terms of edge preservation and relative image error. Conclusions TV-minimization-based iterative image reconstruction methods may be greatly used for low-dose CT imaging of trabecular bones with reduced number of views.