Fusion reconstruction algorithm to ill-posed projection (FRAiPP) for artifacts suppression on X-ray computed tomography

This study aims to address and test a new fusion reconstruction algorithm (FRA) applying to ill-posed projection to suppress noise artifacts and to improve the image quality, by combining iterative algorithm to the filtered backprojection (FBP) for computed tomography (CT). The reconstruction technique based on iterative reconstruction (IR) is used to approximate the value of the ill-posed pixel, then the forward projection operator is employed to obtain a new sinogram as a compensation item, and the superior one is selected to replace the inferior projection which is reconstructed from FBP. By utilizing the flexibility of the tradeoff coefficient in terms of sinogram updating, the new projection was constructed and the artifacts were eliminated/reduced. The study results show an effective suppression in noise artifacts of ill-posed projection. For simulation model # Blade and # Pi, the Normalized Mean Square Distance (NMSD), and Normal Average Absolute Distance (NAAD), with the proposed method were reduced by 70% above, Universal Quality Index (UQI) was increased by 60% above; It also indicates better uniformity of the results to the real projection of # Yueya, where the NMSDs and NAADs are well reduced by 28.82% and 29.34% for our proposed method, and UQIs are well improved about 65.96%. Based on the global noise suppression of the IR and the accuracy of the FBP algorithm, it is encouraged by the promising performance of the proposed algorithm to artifact reduction and detail preservation. Studies on simulation phantom and real projections were verified on the CT system. The results demonstrate the potential of our method for medical and industrial imaging.