Room: Exhibit Hall | Forum 2
Purpose: To determine the registration variability in Cone Beam (CB) CT protocols using Varian On-Board Imaging (OBI) and Elekta X-ray Volume Imaging (XVI) software.
Methods: Each available CBCT protocol on the Varian OBI (9) and Elekta XVI (12) systems was acquired on a CIRS ISO Cubeâ„¢ phantom with known 3D translational, 3D rotational, and 6D translation and rotational offsets. Automatic registration was performed, and manually fine-tuned if necessary. Protocol and user variability was determined by the standard deviation of full arc and partial arc protocols repeated five times. Deviations from mean shifts and expected shifts (3D translational only) between different protocols were compared.
Results: Translational protocol and user variability was determined to be <0.03cm and <0.05cm for Varian OBI and Elekta XVI registrations, respectively. After automatic registration, 4/27 OBI and 8/36 XVI registrations required manual adjustments and final registrations were used for analysis. Deviations from expected translational shifts for OBI and XVI registrations were systematic in all three directions, likely representative of a set-up error larger than the inter-protocol variability. For all protocols, translational and rotational deviations from mean were <0.02cm and <0.2Ëš, respectively for OBI, and <0.08cm and <0.6Ëš, respectively for XVI except for a 1.2Ëš roll deviation for the ProstateM15 protocol. Standard deviations in translational and rotational shifts were <0.01cm and <0.1Ëš, respectively, for OBI and <0.4cm and <0.5Ëš, respectively, for XVI.
Conclusion: Based on preliminary results, we find no clinically significant differences in registration results between CBCT protocols for either vendor. There was a difference between the two vendors in both deviation from mean and standard deviation of shifts. Future investigations will include registration of an anthropomorphic phantom, as well as data acquisition on additional treatment machines.