Which CT Set for Lung SBRT Planning: The Effect of Varying Lung Density on Monte Carlo Dose Calculation

Abstract

Free-breathing (FB) and average CTs are common CT images used for SBRT lung planning. The FB CT, a fast 3D scan, consists of random snap shots at different breathing phases for different CT bed positions. The average CT, often created as a mean intensity projection from a slow 4D CT, represents a time-averaged anatomy. While the average CT has been shown to be advantageous for GTV contouring and target localization, both are believed to be equivalent for dose calculation with conventional algorithms (pencil beam, convolution superposition). Since previous studies demonstrated considerable MC dose dependence on varying lung densities, whether in phantoms or between patients with different lung densities, we aimed to study the lung density variation between FB and average CTs, and its effect on MC dose calculation increasingly used in lung SBRT planning. With IRB approval, 10 previously-treated lung SBRT patients were randomly chosen. For each patient, both a FB CT and an 8-phase 4D CT were acquired. The 4D CT was used for contouring an ITV while the FB CT was used for treatment planning with dynamic conformal arcs in clinical treatments. Lungs were contoured on the FB CT, on each phase of the 4D CT, and on an average CT created from the 4D CT. Lung density was compared for each patient among these 10 different data sets. Dose distribution was calculated with MC (iPlan, BrainLab) on FB and average CTs for dose comparison using D5, Dmean and D95 of PTV and GTV, as well as RTOG 0813 protocol compliance. Wilcoxon’s signed rank test was used. For the 10 studied patients, the median (range) of the minimum, mean, and maximum lung density from the 4D CT was 0.26 (0.10∼0.34), 0.31 (0.12∼0.43), 0.35 (0.22∼0.51) g/cc, respectively. Using the mean lung density from the 4D CT as the gold standard, the median (range) deviation of that on the FB and average CT was -0.03 (-0.09∼0.02) and 0.00 (-0.01∼0.01) g/cc, respectively. For the PTV, the median (range) D5, Dmean and D95 percent difference between the FB and average CT dose calculation was -1.9 (-5.1∼1.2%), -0.2 (-1.1∼0.4%) and -1.1 (-4.3∼0.7%), and for the GTV, -1.0 (-2.9∼0.8%), -0.2 (-0.4∼0.3%) and -0.8 (-2.4∼0.5%), respectively. There was no dose index difference that was statistically significant (P< 0.05) for the studied population. RTOG compliance was not different between the two CT sets for any patient. Lung density on the average CT accurately represents the mean density from the 4D CT. Varying density differences were observed between the average and FB CTs, hence leading to varying dosimetric differences in MC dose calculated. However, for the 10 studied patients, the dosimetric differences were not statistically significant. This suggests that negligible difference between FB and average CT is expected for MC calculation, similar to what previously believed true for conventional dose algorithms.