Session Item

To evaluate a novel implementation of a CBCT conversion algorithm for dose calculation implemented in RayStation (Development Version 10B-DTK, RaySearch, Stockholm, Sweden). The algorithm uses deformable registration to correlate grey scale intensity values with an intensity conversion function. The algorithm is able to estimate artifacts and create a CBCT correction map. Additionally, the algorithm uses a stitching technique to simulate missing tissue outside the field of view (FOV) of the CBCT.

CBCTs acquired for ten head and neck and ten gynecological patients were collected and converted using the new algorithm (CBCT_c). A bulk density overriding technique implemented in the same version of RayStation was used for comparison (CBCT_b). The CBCTs and the planning-CT (pCT) were rigidly registered. Clinical treatment plans, which were optimized on the pCT, were recalculated on both the CBCT_c and the CBCT_b. The resulting dose distributions were analyzed utilizing the MICE toolkit (NONPIMedical AB Sweden, Umeå) applying local gamma analysis with 1% dose difference and 1 mm distance to agreement criteria. For both CBCT conversion methods, the pCT was used as ground truth. Four different dose threshold level were used for the analysis: 10%, 30%, 50 % and 90%. The 90% threshold was selected to assess the high dose region around the PTV. A paired student’s t-test was applied to test the differences in gamma pass rates (GPRs) between the CBCT_c and CBCT_b method. A p-value smaller than 0.05 considered statistically significant.

Figure 1 and 2 show box-plots of the GPRs grouped by conversion method and threshold for the head and neck and gynecological cases, respectively. On average, the CBCT_c method showed GPRs higher than 95% for all indications and thresholds. The GPRs for the CBCT_b method were systematically lower compared to the CBCT_c method. These differences were also statistically significant for all test cases and thresholds. The main differences between the dose calculated on the CBCTs and the pCT were found in regions where weight loss occured frequently or at air/tissue interfaces, which were also subject to anatomical variations. For both indications, the stitching technique of the new CBCT_c algorithm provided a reasonable approximation of the missing tissue outside the FOV of the CBCT. Consequently, better agreement could be achieved between the dose calculated on the pCT and the CBCT_c compared to CBCT_b.