(Wednesday, 7/17/2019) 10:00 AM - 10:30 AM
Room: Exhibit Hall | Forum 4
Purpose: To quantify if using an average intensity projection (AIP) scan or a 30% phase (mid-ventilation surrogate, MidV) scan as the reference image affects reproducibility/accuracy of lung cancer patient alignment under free-breathing cone beam computer tomography (CBCT) image guidance.
Methods: AIPs were retrospectively created for 16 lung patients (14 SBRT, 2 conventional) originally planned/treated using the MidV scan as reference. The study included 3-5 CBCTs from each patient. Registrations were performed between the AIP-CBCT and between the MidV-CBCT by 5 individuals (student, medical physics resident, MD resident, medical physicist, and MD). The images were rigidly registered, ITV contours were turned on, and no rotations were allowed to reflect real treatment conditions. A “ground-truth� alignment was also determined by registering the maximum intensity projection (MIP) with pseudo-MIP of each CBCT. This pseudo-MIP was created by adjusting the window/level of the CBCT to brightly show the tumor throughout its various respiration-dependent locations during the acquisition. Additionally, the registrations for AIP-CBCT and MidV-CBCT were repeated 3 times by one individual for intra-observer variability assessment.
Results: The following results are reported for the lateral, anterior-posterior, and superior-inferior directions. The average inter-observer standard deviations (stdevs) in alignment difference from “ground-truth� for AIP-CBCT and MidV-CBCT were 0.5mm/0.7mm/0.9mm and 0.6mm/0.8mm/1.1mm, respectively. The average intra-observer stdevs between the AIP-CBCT and MidV-CBCT were 0.4mm/0.4mm/0.7mm and 0.4mm/0.7mm/0.9mm, respectively. The average differences between “ground-truth� alignment and AIP-CBCT registrations and “ground-truth� alignment and MidV-CBCT registrations were 0.6mm/0.7mm/1.1mm and 0.8mm/1.0mm/1.4mm, respectively. A t-test showed the difference along the anterior-posterior direction for AIP/“ground-truth� and MidV/“ground-truth� registrations differed significantly (p=0.004). By averaging, discrepancies were found to be small; several observations on the effects of patient specific variability were made.
Conclusion: The differences in alignment using AIP versus MidV as the reference images are on average very small; however, individual patients may be impacted more significantly.
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email
