The impact of orthopedic metal artifact reduction software on interreader variability when delineating areas of interest in the head and neck

Purpose: Metal artifacts during computed tomography (CT) hinder the evaluation of diagnostic images and impair the delineation of tumor volume in treatment planning. Several solutions are available to minimize these artifacts. Our objective was to determine the impact of one of those tools on the interreader variability when measuring head and neck structures in the presence of metal artifacts. Methods and materials: Eleven patients were retrospectively selected from an institutional review board-approved study based on the presence of metallic artifacts in the head and neck region. CT raw data were postprocessed using a metal artifact reduction tool. A single matching CT slice from the filtered backprojection and postprocessed data sets was selected in the region of the metal artifact. Areas of selected anatomical structures were measured by independent readers, including an anatomical structure selected from a CT slice with no metal artifact in each patient as control. The intraclass correlation coefficient was calculated. Results: Two extreme outliers were identified and the intraclass correlation coefficient was performed with and without them. The intraclass correlation on filtered backprojection, postprocessed, and control images was 0.903, 0.948, and 0.985 with outliers and 0.884, 0.971, and 0.989 without outliers, respectively, for all readers. On the other hand, the intraclass correlation on filtered backprojection, postprocessed, and control images for experienced readers was 0.904, 0.979, and 0.976 with outliers and 0.934, 0.975, and 0.990 without outliers, respectively. Conclusions: The interreader variability of areas measured in the presence of metal artifact was greatly decreased by the use of the metal artifact reduction tool and almost matched the variability observed in the absence of the metal artifact.