MO‐F‐BRC‐01: 3D Target Trajectory Reconstruction Using CBCT Projection Images

Purpose: With wide use of CBCT in IGRT, extracting the 3D target trajectory from CBCT projection images would be useful to manage intra‐ fraction motion for motion inclusive, gated and tumor tracking delivery methods. The purpose of this work was to develop a target trajectory estimation method from CBCT by employing inter‐dimensional correlation between each direction of target motion, and demonstrate its feasibility with a liver SBRT case. Methods: To estimate the target trajectory from CBCT projection images we utilize an inter‐dimensional correlation between each direction of target motion: LR(t)=a_LR×SI(t)+b_LR, AP(t)=a_AP×SI(t)+b_AP. The parameters of the simple linear correlation model were determined with least‐squares estimation by minimizing the discrepancy between the measured and model‐estimated projection positions on the imager coordinates. The method was demonstrated using a CBCT projection images obtained during liver SBRT treatment for 3D/3D matching purpose using On‐board imager of a Varian iX machine. From each projection image of the CBCT, positions of an implanted gold seed were extracted using Varian research tool (RPM‐Fluoro). The estimated trajectory was compared with the marker trajectory obtained from 10‐binned respiratory phase 4DCT. Results: In general the estimated trajectory followed well the reference trajectory from 4DCT in the case. Mean root‐ mean‐square error of the model estimation was 0.67mm in lateral direction and 0.00mm in longitudinal direction on the imager coordinates. Conclusions: A simple 3D marker trajectory estimation method using CBCT projection images was developed. Preliminary results show that it could be used to accurately localize a moving tumor.