Impact of Small Metallic Artifact Contouring on Local Dose Perturbations
Abstract
Purpose
Small metallic objects such as titanium and tantalum surgical clips are frequently present in clinical treatment planning images. Because these objects produce streak artifacts in planning CT images and may lie within the beam path, there is concern that they could alter calculated dose distributions proximal to the artifact. The purpose of this work was to investigate whether contouring and density override choices for small metallic artifacts meaningfully affect local dose predictions.
Methods
Photon beam measurements were performed in a PMMA phantom containing small titanium rods and tantalum surgical clips positioned along the beam axis, with Gafchromic film placed within the phantom. Acuros XB (AXB) dose calculations were performed using three clinically reasonable contouring and density-override strategies for the metallic inserts: (1) Eclipse automatic segmentation, (2) an institutional auto-segmentation approach, and (3) assignment of the insert region to the surrounding material Hounsfield unit. Film and TPS depth-dose profiles were spatially aligned and compared directly, with residuals (film − TPS) evaluated as a function of depth, focusing on regions immediately upstream and downstream of the metal objects.
Results
Dose differences between film and AXB were confined to the immediate vicinity of the metallic inserts, with the Eclipse auto-segmentation approach producing the largest near-interface deviations relative to the institutional auto-segmentation and uniform HU assignment strategies. Beyond approximately 1–2 cm downstream of the metal objects, residuals for all three contouring approaches were small and approached zero, indicating minimal sensitivity of downstream dose to contouring or density-override choice.
Conclusion
For the small metallic artifacts evaluated, differences in contouring and material assignment produced localized dose variations near the metal interfaces but did not lead to clinically meaningful downstream dose differences. These findings indicate that, for small metallic objects, downstream AXB dose calculations in Eclipse are relatively insensitive to reasonable contouring and density-override strategies.