Till Örebro universitet

Purpose: To evaluate for proton therapy treatment planning the feasibility of two commercial metal artifact reduction (MAR) algorithms in CT-imaging.

Materials and Methods: A head phantom with removable dental fillings and a body phantom with a removable hip prosthesis were scanned to evaluate O-MAR (Philips) and iMAR (Siemens). Reference images (scans without metal) were acquired and subtracted from the uncorrected (no MAR) and MAR-images. CT number-differences were mapped to differences in stopping power ratios to water. In addition, proton treatment plans for a parotid, tonsil and prostate-target were optimized based on uncorrected and MAR images and recalculated on reference images. Beams were arranged to not traverse metal, enabling evaluation of metal artifact impact on target coverage.

Results: MAR algorithms reduced the most extreme dental filling artifacts, but residual artifacts still remained. iMAR reduced hip prosthesis artifacts to large extent, while considerable artifacts still were present with O-MAR. For parotid and tonsil-plans, D98%to PTV was nearly intact in the reference recalculations for both uncorrected and MAR-based plans, with maximum-difference,0.3%. For uncorrected prostate plans, D98%decreased more than 4% in the reference recalculation. For the iMAR prostate plan, D98%was almost identical in the reference recalculation (97.5% versus 97.4%). A slight D98%-decrease was seen in the reference for the O-MAR based plan (96.8% versus 97.5%).

Conclusion: Hip prosthesis artifacts reduced target coverage accuracy, but it was substantially improved with MAR algorithms. Dental filling artifacts were moderately reduced with MAR, but did not substantially affect target coverage