Room: Exhibit Hall | Forum 8
Purpose: Recently the Gamma Index of registered CBCT and CT, combining mass density and distance-to-agreement (DTA), has been used as an effective means to evaluate the quality of image guided pretreatment setup, and to identify radiation-induced patient anatomy change. The pass/fail of the Gamma analysis relies heavily on the Gamma criteria. We propose a regression model based on multiple patient data to quantitatively determine the optimal HU and DTA criteria for CBCT Gamma analysis.
Methods: We retrospectively analyzed daily setup kV-CBCT (acquired using OBI on TrueBeam, Varian Medical) from 10 H&N, 10 thoracic, and 10 abdominal cancer patients. The registration between CBCT and the planning CT was conducted online by therapists and reviewed by radiation oncologist. Visible region of interests were contoured in the CBCT by experts as the ground truth of CT/CBCT similarity. Gamma analyses using different levels of criteria, combinations of mass density 0.2g/cc, 01g/cc, and 0.05g/cc, and DTA 3mm, 2mm, and 1mm, were repeated on CBCT using the MobiusCB (by Mobius Medical System) software. The Gamma passing rate of the total irradiated volume and selected region of interests at different levels of gamma criteria were used as training data for multi-variated parametric regression model to determine the optimal gamma criteria.
Results: Based on the preliminary data, the optimal Gamma criteria is 0.5g/cc and 3mm for head&neck patients, 0.2g/cc and 2mm for thoracic cancer patients, and 0.1g/cc and 2mm for abdominal cancer patients, depending on clinically used setup margins.
Conclusion: Gamma Index of the target volume should be used for CBCT gamma analysis. For different part of the body the optimal CBCT Gamma criteria varies. By using the optimal Gamma criteria, we expect the 90% and 85% Gamma passing rate which are the accepting and warning threshold respectively can be used to accurately guide clinical decisions.
IM/TH- Image Analysis (Single modality or Multi-modality): Computer-aided decision support systems (detection, diagnosis, risk prediction, staging, treatment response assessment/monitoring, prognosis prediction)