Room: Exhibit Hall | Forum 5
Purpose: This study evaluates two methods for correcting CBCT intensities to Hounsfield Units (HU) in order to calculate dose directly on daily CBCT images, an approach intended to improve accuracy in daily dose tracking.
Methods: A planning CT (pCT), a replan CT (rCT), and a CBCT acquired close in time to the rCT were selected from 17 patients across multiple centers and treatment areas. To create a reference dose, the original plan was transferred to the rCT and dose was recalculated on the rCT using a Monte Carlo based algorithm (MCD). The CBCTs were first corrected for shading artifacts. Then, the CBCT intensity values were adjusted by either modifying the rescale slope and intercept for all voxels in the body (C1 method) or the pCT was deformed to the CBCT, and HU values were bulk-assigned into air, soft tissue, and bone classes (C3 method). The corrected CBCTs were then merged with the pCTs by deforming the anatomy on the pCT to the anatomy on the CBCT in the proximity of the CBCT field-of-view. The original plans were then transferred to the final test images. Dose was calculated on each test image using MCD. Gamma pass rates and DVH statistics were calculated.
Results: The C1 method yielded smaller mean dose differences for all analyzed contour groups. In both the high and low dose regions the gamma pass rates were higher for the C1 method than for the C3 method. However, the C3 results are within one standard deviation of the C1 results.
Conclusion: The C1 method showed slight improvement over the C3 method and would likely allow for better daily dose tracking during radiation therapy. In a future study, we will compare both methods to another which uses a deformed planning CT as a surrogate for the daily image.
Funding Support, Disclosures, and Conflict of Interest: Dane Pittock, Nathan Lamba, Shoshana Ginsburg, Alexandria Kruzer, and Aaron Nelson are employees of MIM Software, Inc.