Room: Exhibit Hall | Forum 1
Purpose: To estimate the delivered dose using daily image guided radiation therapy (IGRT) scans and compare it with the planned dose. Additionally, to evaluate the dependence of the process on the imaging modality and deformable image registration (DIR) algorithm.
Methods: 20 prostate cancer patients were included in the study. All were treated with IMRT or VMAT using CT-on-rails (CTOR) or cone beam CT (CBCT) for daily IGRT, respectively. 453 fractions across those patients were used to calculate the cumulative dose in bladder and rectum. Two different DIR algorithms (intensity-based and contour-based) were used for this process. For the contour-based DIR, the contours of bladder and rectum were manually drawn on each CTOR or CBCT image. Dmean, D1cc and V50Gy were used as clinically relevant dose metrics for the comparisons between plan-delivery, CTOR-CBCT, daily-weekly IGRT, DIR algorithms.
Results: For CTOR, the deviation of Dmean between delivered and planned doses was 1.8Â±1.8Gy for bladder and -2.2Â±6.7Gy (p<0.01) for rectum for the intensity-based DIR and 2.1Â±1.6Gy and -3.8Â±2.3Gy (p<0.01) for the contoured-based DIR, respectively. The values for D1cc were -1.9Â±3.2Gy for bladder and -2.9Â±3.5Gy (p<0.01) for rectum for the intensity-based DIR and 0.0Â±0.1Gy and -2.8Â±2.7Gy (p<0.01) for the contoured-based DIR. For CBCT, the corresponding values for Dmean were 19.5Â±11.4Gy (p<0.01) for bladder and 11.8Â±16.0Gy for rectum for the intensity-based DIR and 10.3Â±18.0Gy and 6.6Â±18.8Gy for the contoured-based DIR. The values for D1cc were -0.6Â±0.6Gy for bladder and -0.3Â±1.0Gy (p<0.01) for rectum for the intensity-based DIR and -0.9Â±0.4Gy (p<0.01) and -2.0Â±2.2Gy for the contoured-based DIR.
Conclusion: The accuracy of dose accumulation can be improved using the contour-based DIR algorithm, which however requires physician-defined OARs on the daily IGRT scans. The use of CTOR helped both DIR algorithms perform better compared to CBCT. Weekly dose accumulation was shown to be an acceptable approximation.