Quantitative Evaluation of Metal Artifacts in CT and Cone-Beam CT Images On the Accuracy of Dose Calculation
S Oyewale*, S Ahmad, I Ali, Oklahoma Univ. Health Science Ctr., Oklahoma City, OKSU-E-T-520 Sunday 3:00:00 PM - 6:00:00 PM Room: Exhibit Hall
Purpose: To investigate effects of metal artifacts on the dose calculation accuracy of three dose calculation algorithms on CT and cone-beam CT images.
Methods: A phantom with and without metal objects was scanned using CT simulator and on-board kV CBCT. Conformal 5-field treatment plans with identical physical and dosimetric parameters were created in the Eclipse™ treatment planning system for CT and CBCT images. The beams were designed to avoid entering through metallic object. Three dose calculation algorithms were used in the plans: (a) pencil beam convolution with modified Batheo (PBC-MB), (b) equivalent TAR (PBC-ET), and (c) analytical anisotropic algorithm (AAA). The dose calculation on the phantom CT images without metal was used as reference to compare with doses calculated using scans with metal objects. The gamma analysis and percentage dose differences were used to quantify the dosimetric effects of image artifacts from metal on CT and CBCT numbers.
Results: The dose difference analysis shows that mean percentage dose differences (MPDD) for CBCT with metal artifacts ranged from -0.12% to -0.26% with an average of -0.20% for PBC-MB, -0.25% to -0.39%, with average -0.31% for PBC-ET, and -0.25 to -0.84%, with average -0.62% for AAA for all planes. The PBC-MB showed the least difference in dose and highest gamma passing rates with the AAA consistently having the highest differences and lowest gamma passing rates. This might result from accurate consideration of scattered dose in heterogeneities by AAA. All MPDD for doses calculated on CT and CBCT using different dose calculation algorithms were not significant (within 1%).
Conclusion: The image artifacts due to streaking from metal objects were generally small (within 1%) for the dose calculated using the different treatment planning algorithms on CT and CBCT images. Furthermore, the dose difference from CT-number variation between CBCT and CT were negligible.