Evaluation of setup accuracy for NSCLC patients; studying the impact of different types of cone-beam CT matches based on whole thorax, columna vertebralis, and GTV

Purpose. The aim of this study is to evaluate the patient setup accuracy by investigating the impact of different types of CBCT matches, performed with 3 (translations only) or 6 (including rotations) degrees-of-freedom (DOF). The purpose is also to calculate and compare CTV to PTV margins based on the various CBCT matches, setups using 2D kV planar imaging or setups using skin markers only (non-IGRT). Material and methods. Setup images from 16 NSCLC patients with weekly CBCT and daily 2D kV planar imaging were analyzed retrospectively. The CBCT matches were based on the columna vertebralis (CV), the whole thorax (WT) and the soft tissue (ST) delineated GTV, where the ST match was chosen as reference. Thus the translational and rotational shifts in three dimensions were assessed. Finally, setup margins were calculated using van Herk's margin recipe. Results. For 80% of the investigated 3 DOF/2D kV CV setups, the translational shifts were within [-3, 2] mm for all three directions. Corresponding values for the 6 DOF/non-IGRT CV and the 6 DOF/non-IGRT ST matches were [-5, 8] mm. Furthermore, 80% of all setups were within +/- 2 degrees for pitch-, roll- and yaw-rotations, and none exceeded 5 degrees. The calculated margins for non-IGRT, about 10 mm, were reduced to approximately 4 mm, regardless of using IGRT setup by CBCT or 2D kV imaging on CV. However, if using WT CBCT setup, the margin in LNG direction was slightly larger, approximately 6 mm. Conclusion. IGRT for NSCLC is an essential tool for margin reduction, since patient setups based on IGRT leads to approximately half the margin sizes compared to non-IGRT setups. Both CBCT and 2D kV planar imaging yields approximately the same margins for CV/ST matches. The magnitudes of the patient rotations were