Background: To explore the feasibility and effectiveness of the metal artifact reduction software (MARs) reconstruction algorithm in reducing metal artifacts of knee prostheses and to explore the optimal monochromatic level of virtual monochromatic spectral (VMS) images for artifact reduction to provide highquality images and reliable diagnosis in patients after total knee arthroplasty (TKA).Methods: A total of 31 patients underwent gemstone spectral computed tomography. VMS images with MARs and without MARs were obtained at different energy levels (80, 100, 120, and 140 keV). Two observers scored each group of images, and interobserver agreement was evaluated. Artificial indices (AIs), percentage500HU and structural similarity index measure (SSIM) values were calculated in the objective analysis to evaluate the image quality and impact of metal artifacts.
Results:The consistency of the scores of the 2 observers was good (kappa value =0.78), and the score of the VMS images with MARs was higher than that of VMS images without MARs. AI values and percentage 500HU of the MARs group were significantly lower than those of the without MARs group, while SSIM values were significantly higher. In the comparison of different keV images, the AI value decreased with the increase in keV in the range of 80-120 keV, but there was no significant difference between the 120 keV images and 140 keV images. In the group with MARs, the percentage 500HU of 100-140 keV images was significantly lower than that of the 80 keV images, but there was no significant difference between 100, 120, and 140 keV images. In the group without MARs, the percentage 500HU was significantly different among all keV groups.Conclusions: VMS images combined with the MARs algorithm can significantly reduce the metal artifacts of knee prostheses and improve image quality. At an energy level of 100-120 keV, a good metal artifact removal effect and soft tissue contrast can be achieved, and the best metal artifact removal effect can be achieved at 140 keV.