Purpose: In the context of lung SBRT where dose gradients are steep and few fractions are used, capturing changes in target motion relative to simulation is imperative. The use of 4D CBCT is likely to better capture these changes than 3D CBCT due to the incorporation of temporal information.
Methods: Nine previous patient waveforms were programmed into a dynamic thorax phantom using a 1 cm target to simulate motion due to respiration. Each waveform was imaged with 4D CT and an SBRT plan was optimized using the ITV from the maximum intensity projection. 5 mm and 10 mm expansions of the ITV contour were created to simulate increased SI motion at the time of treatment. After planning, 3D CBCT and 4D CBCT images were acquired of the waveforms with 5 mm and 10 mm of additional amplitude motion. The target was manually contoured on all datasets individually. Contours were then used to compare 3D CBCT and 4D CBCT datasets with respective expansions of the 4D CT ITV contours as shown in Figure 1. Percent mean dose difference (PMD), difference of area under DVH curve (DVHc) and dice coefficient (DSC) were evaluated.
Results: The median PMD, DVHc and DSC for the 5 mm of added motion were 2.57% and 0.70% (p = 0.01), 2.13 and 1.18 (p = 0.05), and 0.86 and 0.89 (p = 0.003) for 3D CBCT and 4D CBCT respectively. The median PMD, DVHc and DSC for the 10 mm of added motion were 4.03% and 2.10% (p = 0.008), 3.37 and 1.94 (p = 0.008), and 0.83 and 0.88(p = 0.01) for 3D CBCT and 4D CBCT respectively.
Conclusion: Employing 4D CBCT for image guidance for lung SBRT provides a better representation of target motion changes between treatment and simulation when compared to 3D CBCT.
Cone-beam CT, Image Guidance, Lung