SU‐E‐T‐541: Measurement of CT Density Model Variations and the Impact On the Accuracy of Monte Carlo (MC) Dose Calculation in Stereotactic Body Radiation Therapy for Lung Cancer

Abstract

Purpose:To measure the CT density model variations between different CT scanners used for treatment planning and impact on the accuracy of MC dose calculation in lung SBRT.Methods:A Gammex electron density phantom (RMI 465) was scanned on two 64‐slice CT scanners (GE LightSpeed VCT64) and a 16‐slice CT (Philips Brilliance Big Bore CT). All three scanners had been used to acquire CT for CyberKnife lung SBRT treatment planning. To minimize the influences of beam hardening and scatter for improving reproducibility, three scans were acquired with the phantom rotated 120° between scans. The mean CT HU of each density insert, averaged over the three scans, was used to build the CT density models. For 14 patient plans, repeat MC dose calculations were performed by using the scanner‐specific CT density models and compared to a baseline CT density model in the base plans. All dose re‐calculations were done using the same plan beam configurations and MUs. Comparisons of dosimetric parameters included PTV volume covered by prescription dose, mean PTV dose, V5 and V20 for lungs, and the maximum dose to the closest critical organ.Results:Up to 50.7 HU variations in CT density models were observed over the baseline CT density model. For 14 patient plans examined, maximum differences in MC dose re‐calculations were less than 2% in 71.4% of the cases, less than 5% in 85.7% of the cases, and 5–10% for 14.3% of the cases. As all the base plans well exceeded the clinical objectives of target coverage and OAR sparing, none of the observed differences led to clinically significant concerns.Conclusion:Marked variations of CT density models were observed for three different CT scanners. Though the differences can cause up to 5–10% differences in MC dose calculations, it was found that they caused no clinically significant concerns.