Metal artifact reduction in cervix brachytherapy with titanium applicators using dual-energy CT through virtual monoenergetic images and an iterative algorithm: A phantom study

Abolfazl Kanani,Mehran Yazdi,Amir M Owrangi,Sareh Karbasi,Mohammad Amin Mosleh-Shirazi

DOI: https://doi.org/10.1016/j.brachy.2022.07.003

IF: 2.441

Brachytherapy

Abstract:Purpose: To evaluate an iterative metal-artifact reduction (iMAR) algorithm, dual-energy CT (DECT) through virtual monoenergetic images (VMI), and a combination of iMAR and DECT for reducing metal artifact severity (AS) induced by Fletcher titanium applicators used in cervix brachytherapy, the efficacy of which are hitherto unreported. Methods and materials: 120 kVp single-energy CT (SECT) (Siemens) of BEBIG tandem applicators, varying in shape (straight or curved) and diameter (3.5 mm or 5 mm) in a custom-made water-filled phantom, and their DECT images obtained from extrapolation of 80 kVp and 140 kVp, were reconstructed using four methods: DECT through VMI±iMAR, and SECT±iMAR. The DECT images were reconstructed monoenergetically at 70, 150, and 190 keV. AS was evaluated using measured values and statistical analysis. Results: iMAR, DECT, and combined DECT and iMAR reduced AS (p < 0.05). DECT had a lower AS than SECT, even without iMAR (p < 0.025). SECT+iMAR was more effective than DECT-iMAR with VMI at 70 and 190 keV (p < 0.05), whereas showing no statistically significant difference at 150 keV. With DECT and iMAR combined, AS was reduced more effectively compared to the SECT+iMAR or DECT alone. It also reduced the mean interobserver uncertainty by 0.2 mm. Conclusions: These findings indicate that iMAR reduces the AS caused by Fletcher titanium applicators for both SECT and DECT, a combination of iMAR and DECT is superior to either strategy alone, and at low energies, DECT+iMAR also produces similar artifact reduction. These practical strategies promise more accurate source-position and structure definitions in CT-based gynecological brachytherapy treatment planning.