Author: Loren Bell, Charles S Cathcart, Brett William Eckroate, Harold Y Hu, Joshua Kilian-Meneghin, Ke Nie, Fredrick Warburton, Zhenyu Xiong, Keying (Karen) Xu, Ning J. Yue, Yin Zhang, Lili Zhou 👨🔬
Affiliation: RWJBarnabas Health, Basis Scottsdale, Rutgers Cancer Institute of New Jersey 🌍
Purpose: This study compares the dosimetric impact of multiple metal artifact reduction (MAR) algorithms from different CT simulator vendors using an anthropomorphic head phantom for head-and-neck (H/N) radiotherapy planning.
Methods: A RANDO head phantom was modified to simulate metal artifacts by placing a material with density similar to dental amalgam in the right molar region. Four MAR algorithms were evaluated: iMAR-HI (Siemens), iMAR-NC (Siemens), iMAR-DF (Siemens), and OMAR (Philips). A metal-free image set was also acquired, and the dental filling contour was incorporated via image registration to establish the baseline image set. Both CT and dose voxel sizes were 2.5 × 2.5 × 2.0 mm. Contouring and planning followed AAPM TG-119 guidelines for H/N plans, prescribing 5000 cGy to cover 95% of the PTV. After optimizing the plan on the baseline image set, the plan was copied without re-optimization to the Standard (non-MAR) and each MAR-corrected image set. Dose distributions were evaluated by counting the number of voxels with dose discrepancies exceeding ±1%, ±2%, and ±3% and by performing a gamma analysis (1%/1 mm, 10% threshold) between the baseline and each MAR-corrected image set.
Results: Relative to the Standard image set, the number of voxels with dose differences exceeding ±1% between the baseline and the MAR-corrected image sets decreased by 28.7%, 38.5%, 83.1%, and 80.7%, while those exceeding ±2% declined by 57.5%, 73.8%, 98.8%, and 96.3% for iMAR-HI, iMAR-NC, iMAR-DF, and OMAR, respectively. The corresponding gamma pass rates (1%/1 mm, 10% threshold) were 93.3% (Standard), 97.1% (iMAR-HI), 97.7% (iMAR-NC), 99.7% (iMAR-DF), and 98.4% (OMAR).
Conclusion: Although CT metal artifacts have a relatively small impact on H/N dose distributions, they may compromise treatment accuracy. This study indicates that iMAR-DF and OMAR provide superior artifact reduction, underscoring the importance of selecting effective MAR algorithms to maintain dosimetric accuracy in H/N radiotherapy planning.