Impact of Motion Management and Lung Density on Dose Calculation Algorithm Accuracy In Lung SBRT
Abstract
Purpose
To quantify the accuracy of commercial dose calculation algorithms in lung SBRT planning and evaluate how lung density variations resulting from Deep Inspiration Breath Hold (DIBH) versus 4DCT motion management techniques affect dosimetric agreement.
Methods
Twenty-one lung SBRT cases were retrospectively analyzed: 11 treated using DIBH and 10 using 4DCT. Structure volumes and mean Hounsfield Units (HU) for targets and lungs were recorded from MIM (v7.4). Original treatment plans were generated in Eclipse (v16.1) using the Anisotropic Analytical Algorithm (AAA) and subsequently recalculated and reoptimized with Acuros XB (AXB). Dosimetric accuracy was assessed using two independent secondary check solutions: Mobius3D (Collapsed Cone Convolution, CCC) and ClearCalc (Monte Carlo). Evaluation metrics included PTV V100%, PTV D90%, and gamma analysis (2%/2 mm, 10% threshold with 1 mm calculation grid).
Results
Mean lung density was significantly lower in the DIBH cohort (-846 HU) compared to the 4DCT cohort (-753 HU). This density reduction diminished algorithm agreement. AAA plans showed a significant drop in target coverage (V100%) when recalculated with AXB (96.7% vs. 81.1%). Against independent secondary checks, AAA exhibited a mean PTV D90% deviation of -6.4% versus CCC and a gamma pass rate of only 88.9% versus Monte Carlo, with disagreements most pronounced in DIBH cases. Conversely, AXB improved agreement with independent calculations, reducing the mean PTV D90% deviation to -1.5% and boosting mean gamma pass rates > 99.0%. GTV D90% differences remained minimal (< 1.2%) across all algorithms.
Conclusion
Respiratory motion management techniques significantly influence lung density, which directly impacts the accuracy of dose calculation algorithms. The AAA algorithm systematically overestimates dose in low-density lung tissue, an effect amplified by the reduced lung density observed in DIBH. Advanced model-based algorithms, such as Acuros XB or Monte Carlo, are essential for ensuring dosimetric accuracy in lung SBRT, particularly when DIBH is utilized.