Metal artifact reduction in spectral computed tomography for intracavity brachytherapy in cervical cancer patients：a prospective study

Purpose This study aims to evaluate the effectiveness of combining virtual monochromatic images (VMI) with metal artifact reduction (MAR) algorithms in reducing metal artifacts in cervical cancer patients treated with intracavitary brachytherapy. Materials and Methods A total of 50 cervical cancer patients scheduled to undergo intracavitary brachytherapy were included in the study, and their spectral computed tomography (CT) images were acquired. VMI with and without MAR were reconstructed at energy levels ranging from 40 to 140 keV. CT attenuation and image noise (quantified as standard deviation) were measured to calculate the signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR) for the most prominent artifacts, and the artifact index (AI) for the cervix, bladder, and rectum. These objective metrics were compared between the MAR + VMI and no-MAR + VMI groups. Subjective image quality was evaluated using a relative visual grading analysis on a 5-point Likert scale, focusing on artifact reduction, soft tissue contrast, and contouring preference for clinical use. Additionally, the potential geometric distortion introduced by MAR was assessed by measuring the long axis, short axis, and long-to-short axis ratio. Results Quantitative analysis demonstrated that VMI combined with MAR significantly reduced the AI for both the cervix (by 34.74–88.51%) and rectum (by 11.31–81.35%) compared to VMI without MAR across all energy levels (all p  < 0.05). Image quality evaluation revealed that both the SNR and CNR were consistently higher in MAR-corrected images. Specifically, SNR improved significantly at all energy levels, increasing from 0.70–1.00 to 1.75–1.97 (an increase of 95.00–152.11%, all p  < 0.001), while CNR showed significant enhancement from 90 keV onwards ( p  < 0.05). Among all energy levels, VMI with MAR at 140 keV achieved the greatest artifact reduction, evidenced by the highest SNR and CNR and the lowest AI. Subjective assessment further confirmed that 140 keV + MAR offered the best balance between artifact suppression and soft tissue contrast, receiving significantly higher scores for artifact reduction and contouring preference than other energy levels ( p  < 0.001). Distortion analysis showed that with increasing VMI energy, the measured long and short axes of the titanium applicator decreased, and the long-to-short axis ratio approached 1 ( p  < 0.001), suggesting that 140 keV + MAR is optimal for preserving structural accuracy. Conclusions The combination of high-energy VMI with MAR—especially at 140 keV—was identified as the most effective imaging protocol for intracavitary brachytherapy, offering the greatest reduction in metal artifacts, the least geometric distortion, and the highest suitability for clinical contouring.