(Monday, 7/15/2019) 3:45 PM - 4:15 PM
Room: Exhibit Hall | Forum 1
Purpose: Spine metal implant causes artifacts in CT simulation and increases dose calculation uncertainty, which need to be minimized as spinal cord is often very close to lesions and frequently sets dose limits to best cover the target. Metal artifact reduction (MAR) is therefore proposed and implemented for CT simulation. This study investigates the effect of MAR on spine SBRT dose calculation using Monte Carlo (MC) simulation, which is well accepted as the most accurate way for dose calculation with metal implant.
Methods: Spine SBRT patients with titanium (Ti) implant were identified. CTs were acquired with or without MAR using a Siemens CT scanner (SOMATOM Definition AS). VMAT plans were created using Varian’s Eclipse following our institution’s criteria. Each plan was then converted to two geometry models (one with CT_MAR and the other using CT_nonMAR) in our commissioned MC system, MCSIM. Ti implant was assigned with associated cross-section data and physical density. A hybrid voxel size was used (target region: 1mm; elsewhere: 2mm) with ECUT=0.7MeV, PCUT=0.01MV, and a 1-σ dose uncertainty of <1%.
Results: Differences in DVH for PTV and spinal cord were minimal. However, changes of isodose line (IDL) was obvious in artifact region near the metal between CT_MAR vs. CT_nonMAR. A close look at a region of interest in such artifact area shows large dose differences. This may imply that critical organs within significant artifact region near the metal may cause significant dose uncertainty if no MAR was applied.
Conclusion: Although there is minimal difference in DVH for target and spinal cord, changes of IDL may be significant in the artifact area near the metal if no MAR was applied. Critical organs within that region may have a significant dose uncertainty without MAR correction. Further studies are warranted to include more cases and different tumor sites.
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email
