Room: AAPM ePoster Library
Purpose: To perform a quantitative image quality analysis of four major CT vendors’ commercial metal artifact reduction (MAR) solutions and a novel in-house (AMPP) solution, and evaluate their implications on proton dose distributions.
Methods: A H&N anthropomorphic phantom composed of proton tissue equivalent materials, human skull, air cavities and a removable jaw was used. The removable jaw allowed for the exchange of bone equivalent and metal teeth for the creation of a baseline and artifact-filled scans. The phantom was scanned using Philips, Siemens, GE and Toshiba scanners where each metal scan was reconstructed with its respective vendor’s MAR algorithm. Algorithms were evaluated for severity of artifacts (percentage of bad pixels: HU error outside ±20 HU) and CT number accuracy (mean HU number differences and standard deviations inside volumes). Clinically realistic proton treatment plans were designed for a range of target sizes and anatomical locations on the baseline scan. The optimized plans were then copied onto each respective vendor’s MAR scans and the doses were recalculated without reoptimization. DVH and plan quality metrics were evaluated for each.
Results: AMPP outperformed all vendors’ solutions in terms of image quality, showing lower HU differences and fewer bad pixels (4.2% compared to 25.5-65.5%). Dose distributions were negatively impacted by the presence of metal artifacts; the vendor solutions provided varying, but suboptimal, mitigation of this effect. Our in-house algorithm (AMPP) outperformed the vendor’s solutions on all treatment plans for all evaluated metrics, and showed the most comparable DVHs to the baseline (no metal).
Conclusion: Commercial MAR algorithms were ineffective at reducing artifacts in a H&N setting. Correspondingly, they were not successful at mitigating the impact of artifacts on proton dose distributions. Our in-house algorithm outperformed all four commercial vendor solutions in both imaging and dose distributions, and is ready to be implemented on patients.