A comparison study of total variation stokes strategy for low-dose CT image reconstruction

Previous works have demonstrated that the low-dose CT images could be reconstructed by minimizing the total variation (TV) or adaptive weighted total variation (AwTV) of the desired images incorporate data fidelity term. However, due to the piecewise constant assumption of TV model, the reconstructed images are frequently reported to suffer from the undesired patchy artifacts or stair-case effects. In our previous work, we found that these drawbacks could be eliminated by using the total variation stokes-projection onto convex set (TVS-POCS) method for both sparse-view and low-mAs data. In this study, we further validated the benefits of the TVS-POCS method compared to the TV model based adaptive steepest descent-projection onto convex set (ASD-POCS) and AwTV-POCS methods, and both qualitative and quantitative studies were performed using digital Shepp-Logan phantom, Catphan 600 physical phantom and clinical data. The results reveal that the TVS-POCS method can yield images with several noticeable gains, measured by the universal quality index, the full-width-at-half-maximum merit and standard deviation, as compared to its corresponding TV-based algorithms.